Search

Greenhouse production of fruits and vegetables in developed countries is challenged by labour scarcity and high labour costs. Robots offer a good solution for sustainable and cost-effective production. Acquiring accurate spatial information about relevant plant parts is vital for successful robot operation. Robot perception in greenhouses is challenging due to variations in plant appearance, viewpoints, and illumination. This paper proposes a keypoint-detection-based method using data from an RGB-D camera to estimate the 3D pose of peduncle nodes, which provides essential information to harvest the tomato bunches. Specifically, this paper proposes a method that detects four anatomical landmarks in the colour image and then integrates 3D point-cloud information to determine the 3D pose. A comprehensive evaluation was conducted in a commercial greenhouse to gain insight into the performance of different parts of the method. The results showed: (1) high accuracy in object detection, achieving an Average Precision (AP) of AP@0.5=0.96 ; (2) an average Percentage of Detected Joints (PDJ) of the keypoints of PhDJ@0.2 = 94.31%; and (3) 3D pose estimation accuracy with mean absolute errors (MAE) of 11o and 10o for the relative upper and lower angles between the peduncle and main stem, respectively. Furthermore, the capability to handle variations in viewpoint was investigated, demonstrating the method was robust to view changes. However, canonical and higher views resulted in slightly higher performance compared to other views. Although tomato was selected as a use case, the proposed method has the potential to be applied to other greenhouse crops, such as pepper, after fine-tuning. • Accurate 3D pose-estimation of peduncle nodes in a commercial greenhouse. • Combination of colour image and 3D point cloud to estimate 3D pose. • Comprehensive evaluation covering object, keypoints, pose and, viewpoints aspects. • Robust to variations caused by view change. • Canonical and higher views are superior while lateral views should be avoided. [ABSTRACT FROM AUTHOR]

The appetite of fish significantly influences aquaculture efficiency and fish welfare. However, accurately assessing fish appetite has posed a challenging problem. Currently, the study of fish feeding behaviour relies primarily on the overall information obtained from images of fish schools, often overlooking the distinctive behavioural traits of individual fish. Analysing the behaviour of individual fish is hindered by challenges such as intraclass variation and cross-occlusion within real aquaculture environments. To address these challenges, this paper introduces a novel method for assessing appetite based on individual fish behaviour. The ByteTrack model was improved to enable stable tracking of each fish within a school under complex conditions. Additionally, this paper employs the spatiotemporal graph convolutional neural network (ST-GCN) to extract the movement characteristics of individual fish, facilitating accurate appetite assessment. The experimental results demonstrate that the proposed method achieves 98.47% accuracy in appetite assessment, surpassing the performance of other state-of-the-art methods. This paper provides a new opportunity and effective means for analysing fish behaviour and appetite in intricate environments. • ·An appetite assessment method based on YOLOv8-ByteTrack and ST-GCN was proposed. • ·The method integrates fish features into spatial-temporal sequences. • ·The method prevents data loss caused by fish school stacking. [ABSTRACT FROM AUTHOR]

• A deep learning model for drug-target interaction prediction based on a knowledge graph and an intent graph is proposed, • Drug-target interactions be modeled from a more granular perspective, • Using relation path-aware aggregation under the GNN-base paradigm to integrate more information through a larger neighborhood region. Knowledge graph intent graph attention mechanism Predicting drug-target interactions (DTIs) plays a crucial role in drug discovery and drug development. Considering the high cost and risk of biological experiments, developing computational approaches to explore the interactions between drugs and targets can effectively reduce the time and cost of drug development. Recently, many methods have made significant progress in predicting DTIs. However, existing approaches still suffer from the high sparsity of DTI datasets and the cold start problem. In this paper, we develop a new model to predict drug-target interactions via a knowledge graph and intent graph named DTKGIN. Our method can effectively capture biological environment information for targets and drugs by mining their associated relations in the knowledge graph and considering drug-target interactions at a fine-grained level in the intent graph. DTKGIN learns the representation of drugs and targets from the knowledge graph and the intent graph. Then the probabilities of interactions between drugs and targets are obtained through the inner product of the representation of drugs and targets. Experimental results show that our proposed method outperforms other state-of-the-art methods in 10-fold cross-validation, especially in cold-start experimental settings. Furthermore, the case studies demonstrate the effectiveness of DTKGIN in predicting potential drug-target interactions. The code is available on GitHub: https://github.com/Royluoyi123/DTKGIN. [ABSTRACT FROM AUTHOR]

• Proposed a method (DEEP-EP) for prediction of Epigenetic Protein. • Features are extracted by DDE, DPC, and GAAC. • ERCNN is used for model training. • DEEP-EP secured the best performance. Epigenetic proteins (EP) play a role in the progression of a wide range of diseases, including autoimmune disorders, neurological disorders, and cancer. Recognizing their different functions has prompted researchers to investigate them as potential therapeutic targets and pharmacological targets. This paper proposes a novel deep learning-based model that accurately predicts EP. This study introduces a novel deep learning-based model that accurately predicts EP. Our approach entails generating two distinct datasets for training and evaluating the model. We then use three distinct strategies to transform protein sequences to numerical representations: Dipeptide Deviation from Expected Mean (DDE), Dipeptide Composition (DPC), and Group Amino Acid (GAAC). Following that, we train and compare the performance of four advanced deep learning models algorithms: Ensemble Residual Convolutional Neural Network (ERCNN), Generative Adversarial Network (GAN), Convolutional Neural Network (CNN), and Gated Recurrent Unit (GRU). The DDE encoding combined with the ERCNN model demonstrates the best performance on both datasets. This study demonstrates deep learning's potential for precisely predicting EP, which can considerably accelerate research and streamline drug discovery efforts. This analytical method has the potential to find new therapeutic targets and advance our understanding of EP activities in disease. [ABSTRACT FROM AUTHOR]

• We proposed a novel computational method for detect RNA m6A sites. • The proposed method is of powerful ability to predict RNA m6A sites in multiple species or tissues, and of across species or tissue ability. • A user-friendly webserver facilitates recognition of RNA m6A for biologists. N6-methyladenosine (m6A) is the most prevalent, abundant, and conserved internal modification in the eukaryotic messenger RNA (mRNAs) and plays a crucial role in the cellular process. Although more than ten methods were developed for m6A detection over the past decades, there were rooms left to improve the predictive accuracy and the efficiency. In this paper, we proposed an improved method for predicting m6A modification sites, which was based on bi-directional gated recurrent unit (Bi-GRU) and convolutional neural networks (CNN), called Deepm6A-MT. The Deepm6A-MT has two input channels. One is to use an embedding layer followed by the Bi-GRU and then by the CNN, and another is to use one-hot encoding, dinucleotide one-hot encoding, and nucleotide chemical property codes. We trained and evaluated the Deepm6A-MT both by the 5-fold cross-validation and the independent test. The empirical tests showed that the Deepm6A-MT achieved the state of the art performance. In addition, we also conducted the cross-species and the cross-tissues tests to further verify the Deepm6A-MT for effectiveness and efficiency. Finally, for the convenience of academic research, we deployed the Deepm6A-MT to the web server, which is accessed at the URL http://www.biolscience.cn/Deepm6A-MT/. [ABSTRACT FROM AUTHOR]

• In the realm of virtual screening, the choice of databases greatly influences outcomes. • Commercial databases often pose challenges, including patent restrictions, compound toxicity, and adverse side effects. • Addressing this, the study harnesses the power of generative RNNs with LSTM cells to learn drug properties from DrugBank. • The result is a novel virtual screening compounds database comprising 26,316 compounds. • Comprehensive tests validate its quality in terms of chemical space, ADME attributes, fragmentability, and synthesizability. • Notably, this method unveils a set of promising lead compounds with entirely new backbones, underscoring its potential for more efficient and safer drug discovery. The process of virtual screening relies heavily on the databases, but it is disadvantageous to conduct virtual screening based on commercial databases with patent-protected compounds, high compound toxicity and side effects. Therefore, this paper utilizes generative recurrent neural networks (RNN) containing long short-term memory (LSTM) cells to learn the properties of drug compounds in the DrugBank, aiming to obtain a new and virtual screening compounds database with drug-like properties. Ultimately, a compounds database consisting of 26,316 compounds is obtained by this method. To evaluate the potential of this compounds database, a series of tests are performed, including chemical space, ADME properties, compound fragmentation, and synthesizability analysis. As a result, it is proved that the database is equipped with good drug-like properties and a relatively new backbone, its potential in virtual screening is further tested. Finally, a series of seedling compounds with completely new backbones are obtained through docking and binding free energy calculations. [ABSTRACT FROM AUTHOR]

• An independent Edge Feature Extraction Module (EFEM) that combines local edge information and global positional semantic information is designed to enhance the extraction and utilization of edge features, thereby improving the problem of losing edge details in down sampling during feature extraction. • A Residual Fusion Module (RFM) is designed to fuse edge features and multi-scale object features during the upsampling process. The two comple-mentary features assist each other in jointly optimizing the entire model to improve segmentation prediction results. • ERF-TransUNet was validated on the Glas dataset and the CRAG dataset. The experimental results show that the model can effectively reduce boundary segmentation errors in cell image segmentation, achieve better indicators in both the Dice coefficient and Hausdorff distance compared with other models. In recent years, cancer has seriously damaged human health, and the morphological structure of cells serves as an important basis for cancer diagnosis and grading. Automatic cell segmentation based on deep learning has become an important means of computer-aided pathological diagnosis. Aiming at the existing problems of rough segmentation boundaries and inaccurate segmentation in cell image segmentation, this paper designs a cell image segmentation network model (ERF-TransUNet) based on edge feature residual fusion from the perspective of mutual complementarity and constraint between edge features and object features. The model uses a hybrid architecture of CNN and Transformer to extract multi-scale features from cell images, and adds independent edge feature extraction modules and residual fusion modules to enhance the extraction of edge features and their constraints when fusing with cell object features, improving the accuracy of cell contour positioning. Through experiments on two gland cell datasets, CRAG and Glas, and comparing the segmentation effects with current popular deep learning models, the network model proposed in this paper has achieved good performance in both Dice coefficient and Hausdorff distance, which can effectively improve the segmentation effect of cell images. [ABSTRACT FROM AUTHOR]

Molecular optimization, which aims to improve molecular properties by modifying complex molecular structures, is a crucial and challenging task in drug discovery. In recent years, translation models provide a promising way to transform low-property molecules to high-property molecules, which enables molecular optimization to achieve remarkable progress. However, most existing models require matched molecular pairs, which are prone to be limited by the datasets. Although some models do not require matched molecular pairs, their performance is usually sacrificed due to the lack of useful supervising information. To address this issue, a domain-label-guided translation model is proposed in this paper, namely DLTM. In the model, the domain label information of molecules is exploited as a control condition to obtain different embedding representations, enabling the model to generate diverse molecules. Besides, the model adopts a classifier network to identify the property categories of transformed molecules, guiding the model to generate molecules with desired properties. The performance of DLTM is verified on two optimization tasks, namely the quantitative estimation of drug-likeness and penalized logP. Experimental results show that the proposed DLTM is superior to the compared baseline models. • The proposed model exploits domain label information as a control condition to generate desired molecules. • A feature fusion module is used to enrich the feature representation of molecules. • A classifier network is designed to identify the categories of generated molecules. • Extensive experiments on the QED and PlogP tasks verify the effectiveness of the proposed model. [ABSTRACT FROM AUTHOR]

Currently, deep learning has achieved remarkable success in estimating plant disease from unmanned aerial system (UAS) images. However, two critical challenges remain unexplored: spatiotemporal variations in disease symptoms and the domain shift between source and target datasets. To overcome these challenges, this paper proposes an approach that incorporates temporal aspects of disease progression using time series analysis. Spatiotemporal information is integrated by combining convolutional neural networks and bidirectional long-short term memory (CNN-BiLSTM) to classify the disease into five severity levels. Various feature extraction methods, including both handcrafted and CNN-based feature extractors, are evaluated. Furthermore, to tackle the problem of domain shift, a feature-level domain adaptation method is proposed. This method aims to learn transferable feature representations that remain consistent despite variations between source and target datasets. This approach enhances the spatiotemporal transferability of the CNN-BiLSTM model, enabling the effective utilisation of historical datasets. The study demonstrates that the CNN-BiLSTM model outperforms traditional time-independent machine-learning methods that rely on handcrafted features. Specifically, the Resnet101-BiLSTM model achieves the highest overall classification accuracy of 89.7% among all tested models evaluated on a one-year dataset. Moreover, it shows superior generalisation with 72.7% accuracy for cross-spatiotemporal disease severity classification using domain adaptation, as demonstrated through two-year experiments. By reducing the domain shift of source and target datasets and harnessing time series high-resolution images obtained throughout the crop growing season, this hybrid approach has substantial potential to advance the assessment of crop disease severity in field conditions. • Compared different CNN-based feature extractors for FOV4 symptom detection in UAS images. • Developed a BiLSTM model for temporal analysis in UAS datasets. • Use of domain adaptation enhanced model generalization. • CNN-BiLSTM with CORAL achieved 72.7% accuracy, proving its effectiveness in diverse real-world applications. [ABSTRACT FROM AUTHOR]

Accurately identifying the flowers and fruits of strawberry seedlings in the greenhouse is the key to automated flower and fruit thinning, which can improve efficiency and reduce labour costs in the cultivation. To address the challenges resulting from the small size, similar colour, and overlapping occlusion of strawberry seedling flowers and fruits, this paper proposes a real-time recognition algorithm (Improved YOLO) for accurately identifying them. Firstly, a Swin Transformer prediction head on the high-resolution feature map of YOLO v7 was constructed to better utilise spatial location information to enhance the detection of small target flowers and fruits, and improve the model's spatial interaction and feature extraction ability in scenes with similar colours and overlapping occlusions. Secondly, the GS-ELAN Optimisation Module for neck of network by GSConv was constructed to suppress shallow noise interference from the high-resolution prediction head and mitigate the increase of parameters resulting from high-resolution prediction heads. The experimental results showed that the Precision(P), Recall(R), and mean Average Precision (mAP) of Improved YOLO are 92.6%, 89.6%, and 92.1%. In the meantime, the Improved YOLO algorithm achieves a frame rate of 45 f/s, satisfying the real-time detection requirements. It is 3.2%, 2.7%, and 4.6% higher than the precision, recall, and mAP of YOLOv7, respectively. The accuracy of this model for detecting flowers and fruits was 93.9% and 91.3%, the recall was 93% and 86.3%, and the average precision was 94.7% and 89.5%, respectively. The Improved YOLO algorithm has a high level of robustness and real-time detection performance, allowing it to quickly and accurately identify the flowers and fruits of strawberry seedlings and provides effective support for the automated management of flower and fruit thinning of strawberry seedlings in greenhouse environments. • High resolution prediction head improves feature extraction for small target. • Swin Transformer is introduced to enhance spatial interaction in complex scenes. • GS-ELAN module is proposed to reduce model parameters and suppress noise. • The proposed method efficiently locates flowers and fruits under different scenes. • A method proposed for detecting fruit and flower accurately before thinning. [ABSTRACT FROM AUTHOR]

The agro-food industry is turning to robots to address the challenge of labour shortage. However, agro-food environments pose difficulties for robots due to high variation and occlusions. In the presence of these challenges, accurate world models, with information about object location, shape, and properties, are crucial for robots to perform tasks accurately. Building such models is challenging due to the complex and unique nature of agro-food environments, and errors in the model can lead to task execution issues. In this paper, MinkSORT, a novel method for generating tracking features using a 3D sparse convolutional network in a deepSORT-like approach, is proposed to improve the accuracy of world models in agro-food environments. MinkSORT was evaluated using real-world data collected in a tomato greenhouse, where it significantly improved the performance of a baseline model that tracks tomato positions in 3D using a Kalman filter and Mahalanobis distance. MinkSORT improved the HOTA from 42.8% to 44.77%, the association accuracy from 32.55% to 35.55%, and the MOTA from 57.63% to 58.81%. Different contrastive loss functions for training MinkSORT were also evaluated, and it was demonstrated that it leads to improved performance in terms of three separate precision and recall detection outcomes. The proposed method improves world model accuracy, enabling robots to perform tasks such as harvesting and plant maintenance with greater efficiency and accuracy, which is essential for meeting the growing demand for food in a sustainable manner. • MinkSORT: A novel method of using 3D sparse convolutions for multi-object tracking. • A novel implementation of the concept of DeepSORT for 3D data like point clouds. • Real world experiments in a challenging tomato greenhouse environment. • An evaluation of MinkSORT showing a tracking accuracy increase up to 7.72 points. [ABSTRACT FROM AUTHOR]

• We considered both the string and graph modal information in the drug and target. • We captured the positional information of string modal to extract more useful information in SMILES and target sequences. • We used multimodal fusion methods to balance and integrate the information from multiple modalities. • We conducted experiments based on two benchmark datasets Davis and KIBA, and FMDTA outperformed the state-of-the-art models. In the biomedical field, the efficacy of most drugs is demonstrated by their interactions with targets, meanwhile, accurate prediction of the strength of drug-target binding is extremely important for drug development efforts. Traditional bioassay-based drug-target binding affinity (DTA) prediction methods cannot meet the needs of drug R&D in the era of big data. Recent years we have witnessed significant success on deep learning-based models for drug-target binding affinity prediction task. However, these models only considered a single modality of drug and target information, and some valuable information was not fully utilized. In fact, the information of different modalities of drug and target can complement each other, and more valuable information can be obtained by fusing the information of different modalities. In this paper, we introduce a multimodal information fusion model for DTA prediction that is called FMDTA, which fully considers drug/target information in both string and graph modalities and balances the feature representations of different modalities by a contrastive learning approach. In addition, we exploited the alignment information of drug atoms and target residues to capture the positional information of string patterns, which can extract more useful feature information in SMILES and target sequences. Experimental results on two benchmark datasets show that FMDTA outperforms the state-of-the-art model, demonstrating the feasibility and excellent feature capture capability of FMDTA. The code of FMDTA and the data are available at: https://github.com/bestdoubleLin/FMDTA. [ABSTRACT FROM AUTHOR]

The establishment of a bio-based circular economy is imperative in tackling the climate crisis and advancing sustainable development. In this realm, the creation of microbial cell factories is central to generating a variety of chemicals and materials. The design of metabolic pathways is crucial in shaping these microbial cell factories, especially when it comes to producing chemicals with yet-to-be-discovered biosynthetic routes. To aid in navigating the complexities of chemical and metabolic domains, computer-supported tools for metabolic pathway design have emerged. In this paper, we evaluate how digital strategies can be employed for pathway prediction and enzyme discovery. Additionally, we touch upon the recent strides made in using deep learning techniques for metabolic pathway prediction. These computational tools and strategies streamline the design of metabolic pathways, facilitating the development of microbial cell factories. Leveraging the capabilities of deep learning in metabolic pathway design is profoundly promising, potentially hastening the advent of a bio-based circular economy. [Display omitted] • Various computational methodologies have emerged for metabolic pathway design. • Deep learning facilitates enzyme discovery in pathway design. • Deep learning approaches in single- and multi-step pathway prediction are reviewed. • Deep learning techniques enhance the construction of microbial cell factories. • Computational guidelines are offered for experimental researchers. [ABSTRACT FROM AUTHOR]

Object detection technology is significant for automatic fish monitoring and intelligent aquaculture. Deep learning fish detection provides great convenience for aquaculture with its unique advantages. However, it still faces several challenges in practical applications. For example, due to the perspective projection effect, the different scales of non-rigid fish will cause difficulties in the identification and location. Also, when labelled data collection is expensive or not feasible, domain shifts due to differences in data distribution can severely limit the practical deployment of algorithms in aquaculture. Hence, this paper proposes a decoupled multi-scale discriminant model especially for cross-domain fish detection to solve these problems, termed DMDnet. The model is divided into three components: multi-scale feature enhancement detector, category adaptive module, and adaptive regression module. An enhanced feature pyramid is embedded into the detector to alleviate the multi-scale problem of fish and improve the discrimination of the whole model. Category and regression adaptors with independent parameters are also introduced to avoid the damage of adversarial training on the detector's discriminability. These two adaptors are separated from the detector, to improve transferability. The cross-domain experiments were carried out on underwater fish images collected from various scenes and aquaculture conditions. The results verify that this method can improve the detector's generalisation performance significantly in the unlabelled new domain. Hence, this method can reduce the cost and increase the efficiency of aquaculture, which has a better application prospect. • Decoupled domain adaptation model proposed for underwater fish detection. • Enhanced feature pyramid introduced to solve the multi-scale of the fish. • Adaptors decoupled from detector to avoid interference between different tasks. • Method can improve generalisation of cross-domain detection in aquaculture. • Cross-domain experiments performed to verify the method and prove its validity. [ABSTRACT FROM AUTHOR]

Mechanical weeding can solve the problems of environmental pollution caused by chemical weeding and high labour intensity of manual weeding. However, the traditional mechanical weeding method will result in the seedling damage due to the bending of the seedling rows. In this paper, an identification method of straight-curved rice seedling rows based on initial clustering at the bottom, and exterior point elimination at the top, is proposed. The approach proposed can effectively solve the influences of several duckweeds, the random weed distribution, changes in light intensity, and the curvature changes of the seedling rows. The identification of straight-curved seedling rows mainly includes the dynamic collection of rice seedling images, the detection of rice seedlings based on the improved CS-YOLOv5 model, and the extraction of seedling strip lines based on initial clustering and exterior point elimination at the top. The field experiments indicated that the average identification angle deviation was less than 1°. • Modifiy a deep learning model to detect rice seedlings under different conditions. • Extract the seedling strip lines of straight-curved rice seedling rows. • Seeding row identification has an average angle deviation of less than 1°. [ABSTRACT FROM AUTHOR]

• We develop a novel model to predict the applicability of immunotherapy for cancer. • We propose a novel method to process unlabeled Anti PD-1 immunotherapy data set. • We analyze the current state and prospects of cancer immunotherapy research. According to global and Chinese cancer statistics, lung cancer is the second most common cancer globally with the highest mortality rate and a severe threat to human life and health. In recent years, immunotherapy has made significant breakthroughs in the treatment of cancer patients. However, only 30% of patients are applicable and may have immune-related adverse events. The traditional immunological inspection methods have limitations and often can not obtain the expected benefits. Deep learning is a typical representation learning method that can spontaneously mine the hidden feature of effective classification from seas of data. In order to alleviate medical resources and reduce costs, this paper proposes a deep learning-based method to predict patients best suited for immune checkpoint blocking therapy from patients CT images. The deep immunotherapy analysis method proposed in this paper is divided into three steps:(1) Using LUNA16 public dataset to develop a deep learning model for nodule detection. (2) Nodule detection was performed on the Anti-PD-1_Lung dataset, and the effectiveness of immunotherapy was determined by comparing the detection results of nodules before and after immunotherapy. (3) After the data set was processed, the deep learning method trained and analyzed the Lung images. According to the experimental results and comparative analysis, the proposed deep immunotherapy analysis method has a good performance in the detection of nodules. It works for the predictions for the applicability of immunotherapy for lung cancer. 1 1 https://github.com/Giat995/Applicability-of-Immunotherapy.git. [ABSTRACT FROM AUTHOR]

• We develop a novel RNA secondary structure prediction algorithm based on the Conditional Deep Convolutional Generative Adversarial Network (CDCGAN) model. • MSFF-CDCGAN proposes a novel image-like representation of RNA structures to expand the original RNA data set. • MSFF-CDCGAN can predict long-sequence RNAs and pseudoknots. Access to RNA secondary structure is a prerequisite for understanding and mastering RNA function. RNA secondary structures play an important role in cells, they can cause or contribute to neurological disorders and can be applied in the medical field. However, the experimental method to obtain RNA secondary structure is costly, laborious and not universal. Although computational methods can predict RNA secondary structure more accurately for short-sequence RNAs, it cannot predict long-sequence RNAs and pseudoknot, which is the bottleneck of RNA secondary structure prediction at present. In recent years, researchers have attempted to use deep learning algorithms to predict RNA secondary structure and have achieved results. However, the small amount of data on the secondary structure of long-sequence RNAs leads to the low accuracy of deep learning methods to predict the secondary structure of RNAs across races. Similarly, RNA structure with pseudoknot is very complex and insufficient data caused the deep learning algorithm to struggle to predict the secondary structure of RNA containing pseudoknots. The RNA data are encoded into grayscale images by a unique encoding method based on the real RNA secondary structure and sequence information. Then, this paper reasonably expands the image data to increase the amount of RNA data, solves the problem of insufficient data for predicting long sequences and RNA secondary structure with pseudoknots in current deep learning methods, and provides a good data foundation for deep learning.The article proposes a multi-scale feature fusion Conditional Deep Convolutional Generative Adversarial Network prediction model (MSFF-CDCGAN) based on the improved Conditional Deep Convolutional Generative Adversarial Network (CDCGAN) model to predict RNA secondary structure. The experimental results showed that the MSFF-CDCGAN model could predict long-sequence RNAs and pseudoknots more accurately than traditional prediction methods. This paper introduces Generative Adversarial Network (GAN) to RNA secondary structure prediction for the first time. It uses a unique image encoding approach to expand the original RNA data set, thus transforming the structure prediction problem into an image analysis problem and effectively solving the bottleneck in RNA secondary structure prediction. [ABSTRACT FROM AUTHOR]

This paper introduces a novel machine learning-based optimization strategy for multi-functional acoustic black hole (ABH) metaplates. The primary objective is to achieve a multi-functional metaplate with excellent performance in elastic wave attenuation and load-bearing capacity simultaneously. The paper begins by describing the design of nanocomposite ABH metaplates, presenting a new pathway to realize multi-functional metaplates. Then, a semi-analytical method, based on plate theory and the Bloch–Floquet theorem, is introduced to consider the band structure of the nanocomposite metaplates. Through systematic analysis, the impacts of the ABH effect, nanocomposite reinforcements, and the viscoelastic damping layer on the bandgaps and strain energy compliance are highlighted. Meanwhile, two optimization objectives representing bandgap characteristics and in-plane stiffness are derived respectively. Subsequently, a deep learning surrogate model is employed to establish a relationship involving significant parameters with the optimization objectives. The performance evaluation confirms accuracy and computational speed of the surrogate model. Finally, an optimization strategy based on deep reinforcement learning is proposed to obtain multi-functional metaplates with superior bandgaps, enhanced in-plane stiffness, or both. The robustness and efficiency of the strategy are demonstrated under different tests. The results show that the proposed strategy can achieve identical results as the genetic algorithm and nondominated sorting genetic algorithm-II, while surpassing them in computational efficiency and balancing multiple objectives. The findings of this study serve as valuable references for the future development and application of multi-functional advanced metamaterials. [ABSTRACT FROM AUTHOR]

The scattering transform is a multilayered, wavelet-based transform initially introduced as a mathematical model of convolutional neural networks (CNNs) that has played a foundational role in our understanding of these networks' stability and invariance properties. In subsequent years, there has been widespread interest in extending the success of CNNs to data sets with non-Euclidean structure, such as graphs and manifolds, leading to the emerging field of geometric deep learning. In order to improve our understanding of the architectures used in this new field, several papers have proposed generalizations of the scattering transform for non-Euclidean data structures such as undirected graphs and compact Riemannian manifolds without boundary. Analogous to the original scattering transform, these works prove that these variants of the scattering transform have desirable stability and invariance properties and aim to improve our understanding of the neural networks used in geometric deep learning. In this paper, we introduce a general, unified model for geometric scattering on measure spaces. Our proposed framework includes previous work on compact Riemannian manifolds without boundary and undirected graphs as special cases but also applies to more general settings such as directed graphs, signed graphs, and manifolds with boundary. We propose a new criterion that identifies to which groups a useful representation should be invariant and show that this criterion is sufficient to guarantee that the scattering transform has desirable stability and invariance properties. Additionally, we consider finite measure spaces that are obtained from randomly sampling an unknown manifold. We propose two methods for constructing a data-driven graph on which the associated graph scattering transform approximates the scattering transform on the underlying manifold. Moreover, we use a diffusion-maps based approach to prove quantitative estimates on the rate of convergence of one of these approximations as the number of sample points tends to infinity. Lastly, we showcase the utility of our method on spherical images, a directed graph stochastic block model, and on high-dimensional single-cell data. [ABSTRACT FROM AUTHOR]

Underwater Image Enhancement (UIE) addresses challenges in marine resource exploitation such as absorption, noise, and low contrast. This paper employs bibliometric methods and data from Web of Science to analyze UIE literature. The analysis reveals that research on UIE entered a rapid development stage starting from 2017. Three main areas of technical support for research: underwater image enhancement methods, deep learning-based methods, and underwater image restoration and depth estimation. Hotspots in research include enhancement, deep learning, restoration, feature extraction, and quality assessment. Drawing from the bibliometric analysis, this paper proposes a conceptual framework for UIE research and presents five research suggestions: enhancing method robustness and adaptability, improving real-time performance, integrating software and hardware, establishing underwater image benchmark datasets, and refining quality evaluation systems for underwater images. This study offers valuable references and guidance for the future exploration and development of the UIE field. • Bibliometric analysis of underwater image enhancement (UIE) literature using Web of Science data. • Three main technical areas: enhancement methods, deep learning, restoration/depth estimation. • Research hotspots: enhancement, deep learning, restoration, feature extraction, quality assessment. • Proposes conceptual framework and five suggestions for future UIE research direction. • Recommendations include enhancing robustness, real-time performance, software–hardware integration, benchmark datasets, refined quality evaluation. [ABSTRACT FROM AUTHOR]

Feeding is an important link in aquaculture that not only affects the healthy growth of fish but is also the main factor that determines the cost and economic benefits of aquaculture. At present, feeding in aquaculture is mainly based on mechanical timing and quantity, relying on a fixed amount of feed preset by the breeder, which does not take into account the dynamic feeding needs of the fish. Problems of underfeeding or overfeeding may occur, affecting the growth rate of the fish and polluting the aquaculture water. Machine vision is an efficient, economical, non-destructive, and objective detection and analysis technology, which is of great significance in promoting the automation and intelligence of aquaculture. Therefore, the combination of machine vision and fish feeding demand for feeding is helpful to improve efficiency and achieve intelligent aquaculture. This paper reviews the research on intelligent fish feeding based on machine vision. The main factors affecting feeding are introduced, followed by a general description of the process of intelligent fish feeding based on machine vision, and a detailed overview and analysis of the key technologies involved in image acquisition, image processing and analysis, and intelligent fish feeding. It also discusses the challenges and potential solutions in intelligent feeding. In short, it aims to help researchers and industry practitioners to better understand the state of the art of machine vision in intelligent fish feeding, and to assist in promoting accurate feeding and improving the efficiency of aquaculture. • Recent advances of intelligent fish feeding based on machine vision are summarised. • The key technologies involved in intelligent fish feeding are introduced. • It provides support for promoting accurate feeding and improving efficiency. [ABSTRACT FROM AUTHOR]

The ability to accurately represent and localise relevant objects is essential for robots to carry out tasks effectively. Traditional approaches, where robots simply capture an image, process that image to take an action, and then forget the information, have proven to struggle in the presence of occlusions. Methods using multi-view perception, which have the potential to address some of these problems, require a world model that guides the collection, integration and extraction of information from multiple viewpoints. Furthermore, constructing a generic representation that can be applied in various environments and tasks is a difficult challenge. In this paper, a novel approach for building generic representations in occluded agro-food environments using multi-view perception and 3D multi-object tracking is introduced. The method is based on a detection algorithm that generates partial point clouds for each detected object, followed by a 3D multi-object tracking algorithm that updates the representation over time. The accuracy of the representation was evaluated in a real-world environment, where successful representation and localisation of tomatoes in tomato plants were achieved, despite high levels of occlusion, with the total count of tomatoes estimated with a maximum error of 5.08% and the tomatoes tracked with an accuracy up to 71.47%. Novel tracking metrics were introduced, demonstrating that valuable insight into the errors in localising and representing the fruits can be provided by their use. This approach presents a novel solution for building representations in occluded agro-food environments, demonstrating potential to enable robots to perform tasks effectively in these challenging environments. • A novel method to build world models for robots based on 3D object tracking. • A real world evaluation of our method in highly occluded scenarios. • Detection and association performance decreases with occlusions. • 3D object tracking yields an accuracy up to 71%. • Tracking metrics give more insight than the total count of objects. [ABSTRACT FROM AUTHOR]

This paper studies the approximation capacity of ReLU neural networks with norm constraint on the weights. We prove upper and lower bounds on the approximation error of these networks for smooth function classes. The lower bound is derived through the Rademacher complexity of neural networks, which may be of independent interest. We apply these approximation bounds to analyze the convergences of regression using norm constrained neural networks and distribution estimation by GANs. In particular, we obtain convergence rates for over-parameterized neural networks. It is also shown that GANs can achieve optimal rate of learning probability distributions, when the discriminator is a properly chosen norm constrained neural network. [ABSTRACT FROM AUTHOR]

• Graph matching is fundamental for many clinical applications. • Graph matching serves to find similarities and differences between data acquired at different points in time, different modalities, or different patient data. • This is the first survey paper of graph matching methods for medical imaging. • As many other fields graph matching is moving in the direction of deep learning. The interest on graph matching has not stopped growing since the late seventies. The basic idea of graph matching consists of generating graph representations of different data or structures and compare those representations by searching correspondences between them. There are manifold techniques that have been developed to find those correspondences and the choice of one or another depends on the characteristics of the application of interest. These applications range from pattern recognition (e.g. biometric identification) to signal processing or artificial intelligence. One of the aspects that make graph matching so attractive is its ability to facilitate data analysis, and medical imaging is one of the fields that can benefit from this in a greater extent. The potential of graph matching to find similarities and differences between data acquired at different points in time shows its potential to improve diagnosis, follow-up of human diseases or any other of the clinical scenarios that require comparison between different datasets. In spite of the large amount of papers that were published in this field to the date there is no survey paper of graph matching for clinical applications. This survey aims to fill this gap. [ABSTRACT FROM AUTHOR]

Identification of rumen protozoa in ruminants is significant for species identification, ecological population structure survey, and protozoa behaviour analysis. At present, rumen protozoa identification still needs to be done manually, which is time-consuming and inefficient. To address this issue, this paper proposes a deep learning method for accurate segmentation and recognition of protozoa instances in rumen microscopic images for the first time. Firstly, a microscopic image dataset of protozoa was constructed, which consists of 2671 images of 17 species, 11 genera, and 2 orders, and the images were annotated in detail. Secondly, by comparing two instance segmentation models, an improved YOLACT was presented for efficient protozoan image segmentation. In the proposed method, the SE (Squeeze-and-Excitation) block is integrated into the shallowest layer of the backbone network to enhance the expression of features, and the ReLU activation function of all layers is replaced by FReLU activation function to eliminate the spatial insensitivity of ReLU activation function. The experimental results show that the segmentation accuracy of the improved YOLACT method is 89.45%, which is 0.55% higher than that of the original YOLACT and 2.63% higher than that of the Mask R-CNN method. Meanwhile, the detection speed of YOLACT is 1.7 times that of Mask R-CNN. This work reduces the labour cost of rumen protozoa identification and provides a new method for accurate identification of the microscopic images of protozoa. Besides, the microscopic image dataset of protozoa and the prediction models constructed in this paper will be publicly available. • We use deep neural network to recognise rumen protozoa in microscopic image. • YOLACT outperforms Mask R-CNN in the accuracy of protozoa segmentation. • Improved YOLACT among all methods achieves the best performance. • This work provides a technical reference to the related protozoa research. [ABSTRACT FROM AUTHOR]

Unified Virtual Memory (UVM) relieves the developers from the onus of maintaining complex data structures and explicit data migration by enabling on-demand data movement between CPU memory and GPU memory. However, on-demand paging soon becomes a performance bottleneck of UVM due to the high latency caused by page table walks and data migration over interconnect. Prefetching is considered a promising solution to this problem given its ability to leverage the locality of program memory access patterns. However, existing locality-based prefetching schemes can not handle all the situations. An ideal prefetcher should not only look at narrow regions of the requested address space but also capture global context to deliver a good prediction of the memory access pattern. This paper proposes a novel framework for page prefetching for UVM through deep learning. We first show that a powerful Transformer learning model can provide high accuracy for UVM page prefetching. We then perform analysis to interpret this Transformer model and derive several insights that allow us to design a simpler model to match the unconstrained model's accuracy with orders of magnitude lower cost. We use a pattern-based method to make the UVM page preditor general to different GPU workloads. We evaluate this framework on a set of 11 memory-intensive benchmarks from popular benchmark suites. Our solution outperforms the state-of-the-art (SOTA) UVM framework, improving the performance by 10.89%, improving the device memory page hit rate by 16.98% (89.02% vs. 76.10% for prior art), and reducing the CPU-GPU interconnect traffic by 11.05%. According to our proposed unified metric, which combines the accuracy, coverage, and page hit rate, our solution is approaching the ideal prefetching scheme more than the SOTA design (0.90 vs. 0.85, with the perfect prefetcher of 1.0). • The first framework for CPU-GPU UVM page prefetching using deep learning. • Revealing insights about UVM prefetching to derive lightweight page predictors. • Significantly improves IPC and memory page hit rate while reducing PCI-e usage. [ABSTRACT FROM AUTHOR]

• Encoding DNA sequences by combining the k-mer principle with one-hot coding. • Identifying enhancers and their strength based on one-dimensional convolutional neural network. • Combining models into an ensemble model by the combination strategy of learners in ensemble learning. The enhancer is a DNA sequence that can increase the activity of promoters and thus speed up the frequency of gene transcription. The enhancer plays an essential role in activating gene expression. Currently, gene sequencing technology has been developed for 30 years from the first generation to the third generation, and a variety of biological sequence data have increased significantly every year. Due to the importance of enhancer functions, it is very expensive to identify enhancers through biochemical experiments. Therefore, we need to study new methods for the identification and classification of enhancers. Based on the K-mer principle this study proposed a feature extraction method that others have not used in convolutional neural networks. Then, we combined it with one-hot encoding to build an efficient one-dimensional convolutional neural network ensemble model for predicting enhancers and their strengths. Finally, we used five commonly used classification problem evaluation indicators to compare with the models proposed by other researchers. The model proposed in this paper has a better performance by using the same independent test dataset as other models. [ABSTRACT FROM AUTHOR]

• A damage localization and quantification method without damage labels is proposed. • An efficient method to quantify bridge damage severity level is proposed. • An efficient method for generating damaged structure dynamic responses is proposed. Among the existing unsupervised bridge damage detection methods, it is difficult to quantify bridge damage severity levels, although some of the methods can achieve the localization of bridge damage. In the existing research, the damage index (DI) curve is calculated by the monitoring point dynamic response (inclination or deflection) of the bridge in the damage state, and the localization of the bridge damage can be achieved by the DI curve without the need for damage labels. Based on this bridge damage localization method, this paper proposes an efficient bridge damage severity level quantification method. Since the different DI curves calculated under the same damaged state basically coincide with each other, the DI curves will change with the bridge-damaged state. Therefore, the quantification of the bridge damage severity level is achieved by only calculating DI curves of different damage degrees at the damage location. To efficiently calculate the DI curve, this paper proposes a method for generating the monitoring point dynamic response of the specified damage bridge. By the specified damage bridge finite element model (FEM) static simulation to obtain arbitrary vertical nodal loads and monitoring point responses of the relationship function, and then combined with the healthy state of the bridge monitoring point dynamic response can be generated monitoring point dynamic response of the specified damage bridge. Since the method in this paper is based on real measured data to calculate the DI curve, it is more robust to noise and modeling errors than the FEM dynamic simulation method, and the modeling of this method only needs to update the model stiffness parameter to reduce the modeling cost. In the example of this paper, the method of this paper calculates the DI curve of one damage state only needs about 7 s about 437 times the computational efficiency of the FEM dynamic simulation method, which can achieve the fast quantification of bridge damage severity levels. Numerical simulations demonstrate that the method can localize and quantify multi-damage in bridges under unknown loads, a small number of sensors, noise, and modeling errors. However, the method in this paper still has limitations that do not apply to the localization and quantification of bridge pier damage and nonlinear models. [ABSTRACT FROM AUTHOR]

Deep learning-based speech deception detection research relies on a large amount of labeled data. However, in the process of collecting speech deception detection data, the identification of truth and lies requires researchers to have a professional knowledge reserve, which greatly limits the number of annotated samples. Improving the accuracy of lie detection with insufficient annotation data is the focus of this study at this stage. In this paper, we propose a semi-supervised high-quality pseudo-label algorithm based on multi-constraint optimization (HQPL-MC) for speech deception detection. Firstly, the algorithm exploits the potential feature information of unlabeled data by using deep auto-encoder networks; secondly, it achieves entropy minimization with the help of the pseudo labeling technique to reduce the class overlap distribution of truth and deception data; finally, it improves the quality of pseudo labels by optimizing the unlabeled loss and reconstruction loss to further enhance the classification performance of the model when the labeled data is insufficient. We recorded an interview-style corpus by ourselves and used it in this paper for the experimental demonstration of the algorithm together with the Columbia/SRI/Colorado(CSC) corpus. The detection performance of the proposed algorithm is better than most state-of-the-art algorithms. • A new semi supervised speech deception detection algorithm. • An interview-type Chinese lie corpus is recorded for speech deception detection. • A data enhancement method suitable for acoustic features of speech lie detection. [ABSTRACT FROM AUTHOR]

In this paper, we are interested in the acceleration of numerical simulations. We focus on a hypersonic planetary reentry problem whose simulation involves coupling fluid dynamics and chemical reactions. Simulating chemical reactions takes most of the computational time but, on the other hand, cannot be avoided to obtain accurate predictions. We face a trade-off between cost-efficiency and accuracy: the numerical scheme has to be sufficiently efficient to be used in an operational context but accurate enough to predict the phenomenon faithfully. To tackle this trade-off, we design a hybrid numerical scheme coupling a traditional fluid dynamic solver with a neural network approximating the chemical reactions. We rely on their power in terms of accuracy and dimension reduction when applied in a big data context and on their efficiency stemming from their matrix-vector structure to achieve important acceleration factors (×10 to ×18.6). This paper aims to explain how we design such cost-effective hybrid numerical schemes in practice. Above all, we describe methodologies to ensure accuracy guarantees, allowing us to go beyond traditional surrogate modeling and to use these schemes as references. • Deep Learning-based hybridization speeds up numerical schemes of atmospheric reentry while maintaining high accuracy. • Initializing a scheme with a hybrid code's prediction reduces the convergence time and keeps the exact same guarantees. • Uncertainty analysis provides statistical guarantees concerning approximation errors when using hybridization code. • Neural network approximation error is statistically lower than many other sources of error inherent to numerical simulations. [ABSTRACT FROM AUTHOR]

Plant scientists require high quality phenotypic datasets. Computer-vision based methods can improve the objectiveness and the accuracy of phenotypic measurements. In this paper, we focus on 3D point clouds for measuring plant architecture of cucumber plants, using spectral data and deep learning (DL). More specifically, the focus of this paper is on the segmentation of the point clouds, such that for each point it is known to which plant part (e.g. leaf or stem) it belongs. It was shown that the availability of spectral data can improve the segmentation, with the mean intersection-over-union rising from 0.90 to 0.95. Furthermore, we analysed the effect of uncertainty in the collection of ground truth data. For this purpose, we hand-labelled 264 point clouds of cucumber plants twice and show that the intra-observer variability between those two annotation sets can be as low as 0.49 for difficult classes, while it was 0.99 for the class with the least uncertainty. Adding the second set of hand-labelled data to the training of the network improved the segmentation performance slightly. Finally, we show the improved performance of a 4-class segmentation over an 8-class segmentation, emphasizing the need for a careful design of plant phenotyping experiments. The results presented in this paper contribute to further development of automated phenotyping methods for complex plant traits. • Adding spectral data boosts 3D point cloud segmentation for plant phenotyping. • High intra-observer variability shows that ground-truth collection is challenging. • Labelling the same dataset multiple times can improve the segmentation performance. • Careful design of phenotyping experiments makes a computer scientist's life easier. [ABSTRACT FROM AUTHOR]

The Novel Coronavirus 2019 (COVID-19) is a global pandemic which has a devastating impact. Due to its quick transmission, a prominent challenge in confronting this pandemic is the rapid diagnosis. Currently, the commonly-used diagnosis is the specific molecular tests aided with the medical imaging modalities such as chest X-ray (CXR). However, with the large demand, the diagnoses of CXR are time-consuming and laborious. Deep learning is promising for automatically diagnosing COVID-19 to ease the burden on medical systems. At present, the most applied neural networks are large, which hardly satisfy the rapid yet inexpensive requirements of COVID-19 detection. To reduce huge computation and memory demands, in this paper, we focus on implementing lightweight networks for COVID-19 detection in CXR. Concretely, we first augment data based on clinical visual features of CXR from expertise. Then, according to the fact that all the input data are CXR, we design a targeted four-layer network with either 11 × 11 or 3 × 3 kernels to recognize regional features and detail features. A pruning criterion based on the weights importance is also proposed to further prune the network. Experiments on a public COVID-19 dataset validate the effectiveness and efficiency of the proposed method. [ABSTRACT FROM AUTHOR]

In recent years many amazing deep learning models have been developed, but in the process of practical applications, people often find that these deep learning models have high requirements for hardware storage space and computing power. In Artificial Intelligent of Things (AIoT) scenario, the computing power of the edge or terminal side are relatively limited, therefore, most conventional deep learning models are difficult to be deployed into AIoT devices. It is significant to explore the different performance under different scales of deep learning models. In this paper, we mainly propose a method to analyze the impact of deep learning models with various sizes through various experiments. We employ slimmable network as a Neural Archtecture Search (NAS) tool to realize various model size freely, and evaluate them on the indicators of flops, robustness and accuracy. The experimental results show the variation of flops, robustness and accuracy with the various model sizes, which help understand the impact on performance of deep learning models with different scales in AIoT systems. • Interpreting robustness of deep learning models with different model complexity. • Employing slimmable network to compress neural network. • Understanding the relevance between robustness and accuracy of deep learning models. [ABSTRACT FROM AUTHOR]

• AMBLE is the scheme for heterogeneous devices in federated learning. • AMBLE adjusts local mini-batch and local epoch adaptively. • AMBLE solves the straggler problem caused by system heterogeneity. • AMBLE can add more computation while waiting time caused by stragglers. • AMBLE adopts learning rate scaling to improve performance. As data privacy becomes increasingly important, federated learning applied to the training of deep learning models while ensuring the data privacy of devices is entering the spotlight. Federated learning makes it possible to process all data at once while processing data independently from various devices without collecting distributed local data in a central server. However, there are still challenges to overcome for the system of devices in federated learning such as communication overheads and the heterogeneity of the system. In this paper, we propose the Adjusting Mini-Batch and Local Epoch (AMBLE) approach, which adaptively adjusts the local mini-batch and local epoch size for heterogeneous devices in federated learning and updates the parameters synchronously. With AMBLE, we enhance the computational efficiency by removing stragglers and scaling the local learning rate to improve the model convergence rate and accuracy. We verify that federated learning with AMBLE is a stably trained model with a faster convergence speed and higher accuracy than FedAvg and adaptive batch size scheme for both identically and independently distributed (IID) and non-IID cases. [ABSTRACT FROM AUTHOR]

Malicious traffic detection is one of the most important parts of cyber security. The approaches of using the flow as the detection object are recognized as effective. Benefiting from the development of deep learning techniques, raw traffic can be directly used as a feature to detect malicious traffic. Most existing work usually converts raw traffic into images or long sequences to express a flow and then uses deep learning technology to extract features and classify them, but the generated features contain much redundant or even useless information, especially for encrypted traffic. The packet header field contains most of the packet characteristics except the payload content, and it is also an important element of the flow. In this paper, we only use the fields of the packet header in the raw traffic to construct the characteristic representation of the traffic and propose a novel flow-vector generation approach for malicious traffic detection. The preprocessed header fields are embedded as field vectors, and then a two-layer attention network is used to progressively generate the packet vectors and the flow vector containing context information. The flow vector is regarded as the abstraction of the raw traffic and is used to classify. The experiment results illustrate that the accuracy rate can reach up to 99.48% in the binary classification task and the average of AUC-ROC can reach 0.9988 in the multi-classification task. • We proposed an approach to gradually construct flow vectors from the field vector. • Extract information irrelevant to the payload from the raw traffic as input. • Unique field value representation makes the embedded vector more effective. • The adjustable number of packets in-flow makes the model more flexible. [ABSTRACT FROM AUTHOR]

Communication overhead is the key challenge for distributed training. Gradient compression is a widely used approach to reduce communication traffic. When combined with a parallel communication mechanism method like pipeline, gradient compression technique can greatly alleviate the impact of communication overhead. However, there exist two problems of gradient compression technique to be solved. Firstly, gradient compression brings in extra computation cost, which will delay the next training iteration. Secondly, gradient compression usually leads to a decrease in convergence accuracy. In this paper, we combine parallel mechanism with gradient quantization and periodic full-gradient compensation, and propose a new distributed optimization method named CP-SGD, which can hide the overhead of gradient compression, overlap part of the communication and obtain high convergence accuracy. The local update operation in CP-SGD allows the next iteration to be launched quickly without waiting for the completion of gradient compression and the current communication process. Besides, the accuracy loss caused by gradient compression is solved by k-step correction method introduced in CP-SGD, which provides a gradient correction every k iterations. We prove that CP-SGD has a convergence guarantee and it achieves at least O (1 K + 1 K) convergence rate, where K is the number of iterations. We conduct extensive experiments on MXNet to verify the convergence properties and scaling performance of CP-SGD. Experimental results on a 32-GPU cluster show that convergence accuracy of CP-SGD is close to or even slightly better than that of S-SGD, and its end-to-end time is 30% less than 2-bit gradient compression under a 56Gbps bandwidth environment. In addition, we analyze the performance of CP-SGD when training on 8, 16 and 32 GPUs. It is found that CP-SGD is suitable for most compression-supported update algorithms, and its scalability is approximately linear. • CP-SGD: Distributed Stochastic Gradient Descent with Compression and Periodic Compensation. • CP-SGD help distributed deep learning train with high-accuracy and friendly-communication simultaneously. • Combining the advantages of parallel mechanism and gradient compression. • K-step correction technique makes CP-SGD achieve codesign optimizations and improve compression-based methods greatly. • CP-SGD suits most compression-supported update algorithms and gradient compression strategies. [ABSTRACT FROM AUTHOR]

Due to the relevance of agriculture in economy and human development, the inclusion of technology in this activity is of utmost importance, and moisture content prediction is relevant for assessing the degree of maturity of a crop, which relates to efficient harvesting and quality control. This paper presents an accurate deep learning model for the prediction of the moisture content of canola and wheat crops, based on hyperspectral images taken by several drone flights. This model serves as the starting point for a supervised band selection process that involves a novel approach based on a game-theory model-interpretability analysis. The deep learning model for moisture content prediction included a final ensemble of two branches for analysis of spatial and spectral features, and it reached a coefficient of determination of 0.916 and 0.818 for the canola and wheat test datasets, respectively. SHapley Additive exPlanations analysis allowed us to study the individual predictions of the models, which is the most important contribution of this paper because this approach could eventually lead to the design and implementation of more tailored software and hardware for the analysis of spectral information. The obtained results validate the idea that using this approach actually obtains the spectral bands that are important for this task, since they are similar to PCA results, and they fall on the NIR part of the spectrum, which is widely used in moisture measurement of agricultural products and vegetation analysis. • An accurate model for moisture content prediction, using hyperspectral images. • An ensemble of deep learning models that uses spectral and spatial information. • A novel approach to assess the importance of spectral bands in our predictions. • Results are consistent with current theoretical and experimental knowledge. • This approach could lead to novel tools for analyzing hyperspectral information. [ABSTRACT FROM AUTHOR]

Over the last two decades, research in hyperspectral imaging has been increasing and its use in horticulture is expected to be spreading in the coming years. The emerging techniques are currently gaining interest of the research community. However, there are still challenges to the applicability. In this review we demonstrate that hyperspectral imaging can be used as an effective tool for fruit, vegetables and mushrooms in assessing quality parameters related to well defined variables that can be analysed in the laboratory, as well as complex properties such as maturity, ripeness, detection of biotic defects, physiological disorders, mechanical damages, and sensory quality. Therefore, this paper starts by giving an overview of the quality concept of produce, measuring principles, theory and analysis of hyperspectral imaging systems. Then, emerging techniques to monitor and assess quality parameters, both pre- and postharvest, are described, as well as applications of these are reviewed and discussed. Afterwards, this review proceeds by illustrating the current and potential use of artificial intelligence and its subdomains, machine learning and deep learning, for hyperspectral imaging analysis in horticulture. Lastly, some challenges and considerations for future research are highlighted, including improvement of data availability, possible solutions for an improved integration of artificial intelligence and the transfer of knowledge from research parameters to parameters relevant for industrial stakeholders. • Hyperspectral imaging is an effective tool for in assessing quality parameters. • The most abundantly used wavelengths are 601–850 nm, used in over 50% of studies. • PLSR and SVM are most commonly used models for machine learning. • Deep learning in combination with hyperspectral data is unexplored in horticulture. • Improving availability, AI knowledge transfer and standardisation is recommended. [ABSTRACT FROM AUTHOR]

Edge computing has been considered as a leading paradigm to satisfy the low latency demand for some computation-intensive or data-intensive applications, especially for IIoT applications such as automatic line scheduling of the Internet of Vehicles, time-sensitive supply-chain supervision, and smart control of complex industrial processes. In the edge computing environment, app vendors prefer to cache their app data on edge servers to ensure low latency service. However, it is frequently a challenge in practice, because cache spaces on edge servers are limited and expensive. In view of this challenge, a d eep l earning-based e dge c aching o ptimization method, named DLECO, is proposed to reduce the cost during the cache planning process. In this paper, the edge app data caching problem is formulated as a constrained optimization problem. Then, the specific design of DLECO with a deep learning model is shown, which aims to minimize the overall system cost with lower service latency. The performance of DLECO is analyzed theoretically and experimentally with a collection of data from the real world. The experimental results show its superior performance through comparison with three representative methods. • The edge app data caching (EADC) problem is modeled formally, and the problem is proved to be NP-hard. • The deep learning-based edge caching optimization method, named DLECO, is proposed to solve the EADC problem. • DLECO is analyzed theoretically and evaluated against three representative methods through experiments to show its superior performance. [ABSTRACT FROM AUTHOR]

An important component for the advancement of plant breeding, genetics, and genomics research is the rapid and accurate measurement of phenotypic traits of large plant populations. The phenotypic data that are of interest can be at multiple levels of plant organization including organ-level geometric characteristics as well as the spatial organization of the organs. 3D computer vision enabling 3D geometry acquisition and processing promises to supply fast, automated phenotypic data collection. One important component of the processing pipeline is the segmentation of the plant into its structural components, such as leaves, stems, and flowers. In this paper, a novel 3D point-based deep learning network, namely RoseSegnet, is proposed for segmentation of point clouds of rosebush plants to their organs. The network is equipped with two attention-based modules, one for extracting contextual features at the encoder phase, another for feature propagation at the decoder phase. The network processes regions of points in a hierarchical manner, where at each level, point features are aggregated using attention-based operators. The aggregation is performed by incorporating point relations both within and between the receptive fields, defined by the hierarchical organization of points. RoseSegNet outperforms the widely-used architecture PointNet++ by 4% in terms of MIoU on the publicly available ROSE-X data set. Also, it is demonstrated that introducing local surface features together with the spatial coordinates of each 3D point at the input level boosts the segmentation performance of both networks by 9% in terms of MIoU. • A new 3D point-based deep learning architecture for organ segmentation of plants. • Attention-based modules for aggregation and propagation of features. • Exploiting point interactions within and between regions at multiple scales. • Incorporation of local surface features to the deep learning framework. [ABSTRACT FROM AUTHOR]

• A graph normalization method was proposed to integrate phylogenetic information for metagenomic host phenotype prediction. • The method learned better the structure of phylogenetic trees. • The proposed method enhanced the accuracy of machine learning and deep learning models in different metagenomic datasets. Microorganisms play important roles in our lives especially on metabolism and diseases. Determining the probability of human suffering from specific diseases and the severity of the disease based on microbial genes is the crucial research for understanding the relationship between microbes and diseases. Previous could extract the topological information of phylogenetic trees and integrate them to metagenomic datasets, thus enable classifiers to learn more information in limited datasets and thus improve the performance of the models. In this paper, we proposed a GNPI model to better learn the structure of phylogenetic trees. GNPI maintained the original vector format of metagenomic datasets, while previous research had to change the input form to matrices. The vector-like form of the input data can be easily adopted in the baseline machine learning models and is available for deep learning models. The datasets processed with GNPI help enhance the accuracy of machine learning and deep learning models in three different datasets. GNPI is an interpretable data processing method for host phenotype prediction and other bioinformatics tasks. [ABSTRACT FROM AUTHOR]

• Lightweight end-to-end neural network model --- Tiny-COUNET with state-of-the-art cough detection performance. • A technique enables continuous cough detection by leveraging the always-on active noise cancellation microphones in commodity hearables. • The first work validating the possibility of cough detection based on consumer hearables. Cough event detection is the foundation of any measurement associated with cough, one of the primary symptoms of pulmonary illnesses. This paper proposes HearCough, which enables continuous cough event detection on edge computing hearables, by leveraging always-on active noise cancellation (ANC) microphones in commodity hearables. Specifically, we proposed a lightweight end-to-end neural network model — Tiny-COUNET and its transfer learning based traning method. When evaluated on our acted cough event dataset, Tiny-COUNET achieved equivalent detection performance but required significantly less computational resources and storage space than cutting-edge cough event detection methods. Then we implemented HearCough by quantifying and deploying the pre-trained Tiny-COUNET to a popular micro-controller in consumer hearables. Lastly, we evaluated that HearCough is effective and reliable for continuous cough event detection through a field study with 8 patients. HearCough achieved 2 Hz cough event detection with an accuracy of 90.0% and an F1-score of 89.5% by consuming an additional 5.2 mW power. We envision HearCough as a low-cost add-on for future hearables to enable continuous cough detection and pulmonary health monitoring. [ABSTRACT FROM AUTHOR]

Genotype classification plays a vital role in cultivar evaluation, selection, and production. However, classifying plant genotypes by phenotypes remains an unresolved issue. In this paper, a high-accuracy approach is proposed for plant genotype classification. Based on the Densenet201 and bi-directional Long Short-Term Memory model (bi-directional LSTM), a Densenet201-BLSTM model is given in the approach for classifying various genotypes based on time series of plant images. The growth and development dynamic behaviours and important phenotypes of plants are bi-directionally encoded by the proposed Densenet201-BLSTM to model the complex relationship between phenotypes and genotypes. The accuracy of genotype classification obtained by the proposed DenseNet201-BLSTM model on the test dataset reaches 98.31%. The first attempt is made to classify genotypes of panicoid grain crops. What's more, the proposed genotype classification approach will be useful for the classification of progeny accessions based on their similarity to reference accessions. • A high-accuracy approach of plant genotype classification is proposed. • Genotypes were classified by simultaneously considering spatiotemporal features. • The first attempt is made to classify genotypes of panicoid grain crops. • This study provides the foundation for genotype–phenotype correlation research. [ABSTRACT FROM AUTHOR]

• In this study, we propose the first novel predictor named AAPred-CNN based on deep learning models and embedding techniques for the classification of AAP, attempting to apply deep model into this field and make improvements on the precondition performance. • AAPred-CNN is derived from TextCNN (Text Convolution Neural Network), adpoting multiple convolution channels to extract the local features of the input sequences. • Comparative results show that AAPred-CNN outperforms the state-of-the-art methods in all evaluation metrics and gains a remarkable improvement. • Notably, AAPred-CNN can work well even though only hundreds of training samples are given, which is intuitive. Recently, deep learning techniques have been developed for various bioactive peptide prediction tasks. However, there are only conventional machine learning-based methods for the prediction of anti-angiogenic peptides (AAP), which play an important role in cancer treatment. The main reason why no deep learning method has been involved in this field is that there are too few experimentally validated AAPs to support the training of deep models but researchers have believed that deep learning seriously depends on the amounts of labeled data. In this paper, as a tentative work, we try to predict AAP by constructing different classical deep learning models and propose the first deep convolution neural network-based predictor (AAPred-CNN) for AAP. Contrary to intuition, the experimental results show that deep learning models can achieve superior or comparable performance to the state-of-the-art model, although they are given a few labeled sequences to train. We also decipher the influence of hyper-parameters and training samples on the performance of deep learning models to help understand how the model work. Furthermore, we also visualize the learned embeddings by dimension reduction to increase the model interpretability and reveal the residue propensity of AAP through the statistics of convolutional features for different residues. In summary, this work demonstrates the powerful representation ability of AAPred-CNNfor AAP prediction, further improving the prediction accuracy of AAP. [ABSTRACT FROM AUTHOR]

Accurate processing and analysis of non-human primate (NHP) brain magnetic resonance imaging (MRI) serves an indispensable role in understanding brain evolution, development, aging, and diseases. Despite the accumulation of diverse NHP brain MRI datasets at various developmental stages and from various imaging sites/scanners, existing computational tools designed for human MRI typically perform poor on NHP data, due to huge differences in brain sizes, morphologies, and imaging appearances across species, sites, and ages, highlighting the imperative for NHP-specialized MRI processing tools. To address this issue, in this paper, we present a robust, generic, and fully automated computational pipeline, called non-human primates Brain Extraction and Segmentation Toolbox (nBEST), whose main functionality includes brain extraction, non-cerebrum removal, and tissue segmentation. Building on cutting-edge deep learning techniques by employing lifelong learning to flexibly integrate data from diverse NHP populations and innovatively constructing 3D U-NeXt architecture, nBEST can well handle structural NHP brain MR images from multi-species, multi-site, and multi-developmental-stage (from neonates to the elderly). We extensively validated nBEST based on, to our knowledge, the largest assemblage dataset in NHP brain studies, encompassing 1,469 scans with 11 species (e.g., rhesus macaques, cynomolgus macaques, chimpanzees, marmosets, squirrel monkeys, etc.) from 23 independent datasets. Compared to alternative tools, nBEST outperforms in precision, applicability, robustness, comprehensiveness, and generalizability, greatly benefiting downstream longitudinal, cross-sectional, and cross-species quantitative analyses. We have made nBEST an open-source toolbox (https://github.com/TaoZhong11/nBEST) and we are committed to its continual refinement through lifelong learning with incoming data to greatly contribute to the research field. • We proposed nBEST, a deep-learning toolbox for processing MR images of non-human primate brains. • nBEST leverages lifelong learning and 3D U-NeXt models, performing brain extraction, non-cerebrum removal, and tissue segmentation. • Validated with 1,400+ scans from 11 species, nBEST shows extensive applicabilit. • nBEST is available online with a user manual. [ABSTRACT FROM AUTHOR]

This paper proposes a mesh-free computational framework and machine learning theory for solving elliptic PDEs on unknown manifolds, identified with point clouds, based on diffusion maps (DM) and deep learning. The PDE solver is formulated as a supervised learning task to solve a least-squares regression problem that imposes an algebraic equation approximating a PDE (and boundary conditions if applicable). This algebraic equation involves a graph-Laplacian type matrix obtained via DM asymptotic expansion, which is a consistent estimator of second-order elliptic differential operators. The resulting numerical method is to solve a highly non-convex empirical risk minimization problem subjected to a solution from a hypothesis space of neural networks (NNs). In a well-posed elliptic PDE setting, when the hypothesis space consists of neural networks with either infinite width or depth, we show that the global minimizer of the empirical loss function is a consistent solution in the limit of large training data. When the hypothesis space is a two-layer neural network, we show that for a sufficiently large width, gradient descent can identify a global minimizer of the empirical loss function. Supporting numerical examples demonstrate the convergence of the solutions, ranging from simple manifolds with low and high co-dimensions, to rough surfaces with and without boundaries. We also show that the proposed NN solver can robustly generalize the PDE solution on new data points with generalization errors that are almost identical to the training errors, superseding a Nyström-based interpolation method. [ABSTRACT FROM AUTHOR]

Deep Neural Network (DNN) models offer an attractive alternative to existing acoustic source imaging techniques, such as acoustic beamforming, due to their ever-growing potential with increasing computational power. Source resolution of acoustic beamforming methods is limited at lower frequencies and their source maps may possess sidelobes at higher frequencies. However, acoustic beamforming methods are typically robust over a wide range of simulation and experimental conditions, such as (i) the number of sources present, (ii) source frequency and (ii) extraneous noise sources. The performance of DNN models, when these conditions are varied from their specific design criteria, is yet to be investigated and much work is needed in this area before DNN models can be utilized in experiments, such as wind tunnel tests. Furthermore, few studies have been conducted on experimental validation of DNN models, predominately due to the difficulty of large sets of experimentally obtained data needed for DNN model training and the sensitivity of DNN model performance when any of the aforementioned experimental conditions are varied. In this paper, a series of studies on the robustness of DNN models based on numerical data and experimental data are presented. Numerical DNN (NDNN) models are trained using in-phase and random-phase pressure data generated from six sources over design frequencies from 500 Hz to 20,000 Hz. The robustness of the NDNN models is tested via (1) inclusion of extraneous Gaussian white noise, (2) inclusion of extraneous tonal noise near the design frequency, (3) using source frequencies that slightly differ from the design frequencies and (4) using a number of sources that differs from the design source number. DNN model performance metrics are introduced that present a promising framework for future DNN model studies and bridging the gap between NDNN and experimentally trained DNN models. A preliminary experimental validation was conducted using a single speaker that was systematically placed over a speaker grid to generate training data via acoustic superposition, from which an experimentally trained DNN (EDNN) model was produced. The EDNN model yields exceptional noise source localization capability of the DNN model, revealing a promising start for a more sophisticated EDNN model. • Robustness of a numerical DNN model is analyzed for conditions away from its design criteria. • Various conditions are tested, e.g., source number, source frequency, extraneous noise sources. • The numerical DNN model is trained using both in-phase and random phase sources. • Experimental validation is conducted through an experimentally trained DNN model. • The experimental DNN model yields exceptional noise source localization capability. [ABSTRACT FROM AUTHOR]