Search

TEAM Problem 35, a benchmark problem in the optimization of electrical equipment, is a multi-objective, multi-variable optimization problem involving uniform field optimization. This paper proposes a novel guided optimization algorithm to address this problem and achieve optimal magnetic field regulation. First, the deep learning method is used to predict the magnetic field of the solenoid; second, the generated magnetic field cloud map is screened to extract the magnetic field information of the more optimal solution with uniform magnetic field distribution by calculating the variance of the generated magnetic field map; finally, the magnetic field features are combined with the particle swarm optimization algorithm; specifically, the variance of the calculated magnetic field map is added to the objective function of the optimization algorithm to strengthen the optimization process. This integration improves the convergence speed of the optimization process and ensures the continuity and stability of the obtained Pareto solutions. To optimize the solenoid, this paper considers the uniformity of magnetic field distribution and power loss as the primary optimization objectives. The radius of the coil is chosen as the design variable. The optimization results are compared with those obtained by conventional optimization algorithms, demonstrating superior optimization performance. [ABSTRACT FROM AUTHOR]

This Review explores the transformative impact of artificial intelligence (AI) on the early detection of skin cancer, with a specific focus on melanoma, a potentially lethal form of the disease. Beginning with an overview of traditional diagnostic methods and their limitations, this paper delves into the evolution of AI within dermatology, emphasizing its application in image analysis and pattern recognition. A comprehensive examination of AI algorithms for melanoma detection, including machine learning and deep learning models, is provided. This Review critically assesses the performance metrics, training datasets, and comparative analyses with traditional methods. Addressing challenges such as data quality, interpretability, and ethical considerations, this paper outlines future directions, emphasizing ongoing research, algorithm improvements, and integration with clinical practices. Case studies and success stories highlight the real-world impact of AI in dermatology. This Review concludes by summarizing key findings and underlining the pivotal role of AI in revolutionizing early melanoma detection, with implications for personalized medicine and enhanced patient outcomes. [ABSTRACT FROM AUTHOR]

This paper establishes a data-driven Neural Network (NN) framework. The contact resistance of T2 copper blocks with different roughnesses is predicted by deep learning at room temperature and cyclic loading. The contact resistance problem can be regarded as a regression problem of mapping the high-dimensional array space of multiple variables to the contact resistance. This paper measures the contact resistance of copper blocks with different surface roughnesses under loading and unloading states and obtains the original dataset required by the algorithm. The data characteristics include three surface topography parameters, number of cyclic loads, loading and unloading conditions, and load magnitude, with the data labeled contact resistance. This paper compares the results of the NN model and Holm model results to verify the NN model's effectiveness. The comparison results show that the prediction results of the NN are consistent with the predictions of the Holm model. After training and debugging, the root mean square error of the multiple hidden layers neural network test set is 6.81%, showing a good prediction effect. In conclusion, the deep learning algorithm provides a new way for fast and accurate prediction of the relationship between T2 copper blocks and contact resistance under cyclic loading times and unloading states. [ABSTRACT FROM AUTHOR]

Lenses are fundamental elements in many optical applications. However, various aberrations are inevitably present in lenses, which will affect the distribution of focused light intensity and optical imaging. Accurately predicting the aberrations of a lens is of great significance. Nevertheless, quantitatively measuring the aberrations of a lens, especially when multiple aberrations are present simultaneously, is a challenging task. In this paper, we propose a method based on a designed deep residual network called Y-ResNet to measure the astigmatism and coma of a lens simultaneously. The Y-ResNet was trained on the focused image pattern of a Gaussian beam passing through a lens with astigmatism and coma. The trained network can accurately predict the aberration coefficients of the lens with 0.99 specificity, 0.925 precision, 0.9382 recall, and a 0.9406 F1-score achieved on astigmatism and 0.99 specificity, 0.956 precision, 0.98 recall, and a 0.954 F1-score achieved on coma. Specifically, even if only part of the intensity distribution of the light spot is captured, the network can accurately estimate the aberrations of the lens with an accuracy of over 90% on coma and can identify astigmatism aberration features. This paper can provide a feasible method for correcting beam patterns caused by aberration based on deep learning. [ABSTRACT FROM AUTHOR]

In indoor environments, reverberation often distorts clean speech. Although deep learning-based speech dereverberation approaches have shown much better performance than traditional ones, the inferior speech quality of the dereverberated speech caused by magnitude distortion and limited phase recovery is still a serious problem for practical applications. This paper improves the performance of deep learning-based speech dereverberation from the perspectives of both network design and mapping target optimization. Specifically, on the one hand, a bifurcated-and-fusion network and its guidance loss functions were designed to help reduce the magnitude distortion while enhancing the phase recovery. On the other hand, the time boundary between the early and late reflections in the mapped speech was investigated, so as to make a balance between the reverberation tailing effect and the difficulty of magnitude/phase recovery. Mathematical derivations were provided to show the rationality of the specially designed loss functions. Geometric illustrations were given to explain the importance of preserving early reflections in reducing the difficulty of phase recovery. Ablation study results confirmed the validity of the proposed network topology and the importance of preserving 20 ms early reflections in the mapped speech. Objective and subjective test results showed that the proposed system outperformed other baselines in the speech dereverberation task., (© 2024 Acoustical Society of America.)

Defect detection on wafers holds immense significance in producing micro- and nano-semiconductors. As manufacturing processes grow in complexity, wafer maps may display a mixture of defect types, necessitating the utilization of more intricate deep learning models for effective feature learning. However, sophisticated models come with a demand for substantial computational resources. In this paper, we propose an efficient deep learning framework designed explicitly for mix-type wafer map defect pattern recognition. Our proposed model incorporates several crucial design elements, including lightweight convolutions, bottleneck residual connections, efficient channel attention mechanisms, and optimized activation functions, enabling it to learn spatial and channel features efficiently. We conduct evaluations on a real-world dataset containing 38 mixed-type defect patterns. The experimental results demonstrate that our framework maintains a high level of accuracy while possessing a compact parameter size and remarkably fast inference speed. [ABSTRACT FROM AUTHOR]

The supply of electric power is vital for the daily lives of people, industrial production, and business services. At present, although enough electric power can be supplied to meet the power demand, there are still some challenges, especially in terms of long-distance power transmissions and long-term power storage. Consequently, if the power production capacity exceeds the immediate consumption requirements, i.e., the produced electric power cannot be consumed in a short period, and much electric power could be wasted. Evidently, to minimize the wastage of electric power, it is necessary to properly plan power production by accurately forecasting the future power load. Therefore, a preferable power load forecasting algorithm is crucial for the planning of power production. This paper proposes a novel deep learning model for the purpose of power load forecasting, termed the SSA-CNN-LSTM-ATT model, which combines the CNN-LSTM model with SSA optimization and attention mechanisms. In this model, the CNN module extracts the features from the sequential data, and then the features are passed to the LSTM module for modeling and capturing the long-term dependencies hidden in the sequences. Subsequently, an attention layer is employed to measure the importance of different features. Finally, the output is obtained through a fully connected layer, yielding the forecasting results of the power load. Extensive experiments have been conducted on a real-world dataset, and the metric R2 can reach 0.998, indicating that our proposed model can accurately forecast the power load. [ABSTRACT FROM AUTHOR]

Observing subcellular structural dynamics in living cells has become the goal of super-resolution (SR) fluorescence microscopy. Among typical SRM techniques, structured illumination microscopy (SIM) stands out for its fast imaging speed and low photobleaching. However, 2D-SIM requires nine raw images to obtain a SR image, leading to undesirable artifacts in the fast dynamics of live-cell imaging. In this paper, we propose a single-frame structured illumination microscopy (SF-SIM) method based on deep learning that achieves SR imaging using only a single image modulated by a hexagonal lattice pattern. The SF-SIM method used the prior knowledge to complete the structure enhancement of SR images in the spatial domain and the expansion of the Fourier spectrum through deep learning, achieving the same resolution as conventional 2D-SIM. Temporal resolution is improved nine times, and photobleaching is reduced by 2.4 times compared to conventional 2D-SIM. Based on this, we observed the fast dynamics of multiple subcellular structures and the dynamic interaction of two organelles. The SF-SIM methods provide a powerful tool for live-cell imaging. [ABSTRACT FROM AUTHOR]

Face generation and manipulation techniques based on deep learning have enabled the creation of sophisticated forged facial videos, which are indistinguishable by human eyes. However, the illegal use of deep fake technology will have a serious impact on social stability, personal reputation, and even national security. Therefore, the detection technology of fake facial videos is of great significance to protect national security and maintain social order. Although the existing video-based fake face video detection technology has achieved good detection performance on the public fake face video database, there are still the following problems: (1) the existing technology uses a 2D attention mechanism to obtain local region features from face images and lacks a 3D attention mechanism to obtain local area features from face videos; (2) after obtaining local area features, the existing technology is directly used to classify or only model the inter-regional relationship of images without modeling the temporal relationship between regions of the video. This paper proposes a fake facial video detection framework based on multi-region temporal relationship features, including designing a three-dimensional attention mechanism to extract local features of multiple regions of the face from the video. In order to model the time series relationship between different face areas, a time series graph convolution neural network is also introduced to extract the time series relationship features between multiple areas. In order to model the time sequence relationship between different face regions, the convolution neural network of a time sequence diagram is also introduced to extract the characteristics of the time sequence relationship between multiple regions. Through the change characteristics of the time sequence relationship between face regions, the timing inconsistency of the face video is detected so as to determine whether the face has been deeply forged. Through experiments on multiple datasets, the experimental results of the model accuracy test show that the method proposed in the present invention achieves the highest detection accuracy, and the accuracy in the FaceForensics++ (low definition) dataset is 18.19% higher than that of the benchmark method. The experimental results of the generalization ability test show that the method proposed in the present invention achieves the highest generalization performance, and the detection accuracy in the Celeb-DF dataset is 11.92% higher than that of the benchmark method. [ABSTRACT FROM AUTHOR]

Infrared stealth is an important guarantee for weapon equipment to survive on the battlefield. Emissivity is an important index to measure the infrared stealth characteristics, and the emissivity is closely related to the surface roughness of objects. Therefore, it is an important work to study the relationship between emissivity and roughness. In this paper, the correlation between emissivity and roughness is studied, and the fitting curve and specific relationship are obtained. It is found that the correlation between the emissivity in the 8–14 µm band and roughness is stronger. The cast iron surface roughness dataset is constructed, and a new convolution neural network (CNN) is designed by the feature fusion method, which is the strengthen CNN. The network can effectively extract the detail features in the image, and the model is optimized by the Adam method. Finally, the deep learning model for measuring emissivity based on the optical image is obtained. The effects of different learning rate decay methods, such as piecewise constant decay, exponential decay, cosine annealing, and cosine annealing with warm restart, on the model optimization are studied. The results show that the cosine annealing with warm restart has the best effect, the error of the model is the smallest, and its mean square error is only 0.0014. This paper presents a new idea for the emissivity measurement, which is of great significance to emissivity measurement, infrared stealth, and infrared detection. [ABSTRACT FROM AUTHOR]

Raman spectroscopy, widely used for material analysis, has formed an extensive spectral library. In practical applications, it is usually necessary to preprocess Raman spectroscopy of the target material and then identify the material through spectral-library comparisons. Baseline correction is an important step during pre-processing and it usually requires a special algorithm. However, it demands time and high-level professional skill, confining Raman spectroscopy to laboratories rather than large-scale applications. Therefore, to improve its efficiency and take advantage of the big data in the spectral library, this paper proposes a simple data-augmented deep learning method to achieve Raman spectroscopy recognition without baseline correction. In this method, a simple mathematical baseline (linear or sine function) is added to the spectrum in the database to complete data augmentation. Its training set is used to train the deep learning model. The trained deep-learning model can identify 20 minerals with 100% accuracy without additional baseline correction. Therefore, the method is effective for rapid and direct recognition of Raman spectra. [ABSTRACT FROM AUTHOR]

The reverse design method of acoustic structure based on a deep learning algorithm has been developed as an important means of metamaterial design. In this paper, a multi-band acoustic topological insulator is designed, and the improved competitive search algorithm Long Short-Term Memory (LSTM) algorithm model is used to predict its potential optimal parameter combination to assist the on-demand design of the working frequency band of the multi-band acoustic topology insulator. Finally, the numerical simulation model is established using the optimized structural parameters, and the topologically protected boundary state is studied, which verifies the effectiveness of the method. The research results provide a reference for the on-demand design of multi-band antennas, sound absorption, sound insulation, and other acoustic communication functional devices. [ABSTRACT FROM AUTHOR]

Scanning acoustic microscopy (SAM) is a non-ionizing and label-free imaging modality used to visualize the surface and internal structures of industrial objects and biological specimens. The image of the sample under investigation is created using high-frequency acoustic waves. The frequency of the excitation signals, the signal-to-noise ratio, and the pixel size all play a role in acoustic image resolution. We propose a deep learning-enabled image inpainting for acoustic microscopy in this paper. The method is based on training various generative adversarial networks (GANs) to inpaint holes in the original image and generate a 4× image from it. In this approach, five different types of GAN models are used: AOTGAN, DeepFillv2, Edge-Connect, DMFN, and Hypergraphs image inpainting. The trained model's performance is assessed by calculating the peak signal-to-noise ratio (PSNR) and structural similarity index measure (SSIM) between network-predicted and ground truth images. The Hypergraphs image inpainting model provided an average SSIM of 0.93 for 2× and up to 0.93 for the final 4×, respectively, and a PSNR of 32.33 for 2× and up to 32.20 for the final 4×. The developed SAM and GAN frameworks can be used in a variety of industrial applications, including bio-imaging. [ABSTRACT FROM AUTHOR]

This paper used deep learning algorithms based on the electrical sensing zone (ESZ) method to realize real-time online monitoring and discrimination of particle sorts and their size distribution in liquid–solid systems. Numerical simulations reveal that the characteristics of the temporal pulse signals generated by particles flowing through the ESZ provide the clue to discriminate particle sorts. Based on the above-mentioned results, a deep learning algorithm was used to successfully discriminate and classify micron-sized particles based on the differences in particle pulse signals. This study was conducted to train and validate the residual network (ResNet) model in the deep learning algorithm using four synthetic mixed datasets of different particle pulse signals in a similar size range and to evaluate the performance of the model. Furthermore, the trained ResNet model was applied to the in-house online micron-sized particle analyzer to monitor and discriminate four sets of different particle pulses in a similar size range at the micron level, and the accuracy of the results was above 89%. The results were compared with the measured particle pulse distribution to confirm the feasibility and accuracy of the method. A new method for real-time online monitoring and discrimination of micron-sized particles in industrial production was provided. [ABSTRACT FROM AUTHOR]

Artificial Intelligence (AI) is revolutionizing many aspects of our society, developing a wide variety of real-life applications, from decision-making tasks, such as image classification and autonomous vehicle control, to engineering design, analysis, and manufacturing, such as inverse design. Amin I et al. i [17] wrote an APL Photonics Editor's Pick, demonstrating a novel electro-optic device, which can be used to tune the shape of the neuron nonlinearity based on an ITO-graphene heterojunction integrated absorption modulator in a Si-photonics platform. As a follow-up to a previous APL Photonics special issue on photonics and AI[6] in 2020, the current 2022 APL Photonics special issue discusses the status and future perspectives on photonics and AI in IT, among other topics, highlighting both the role of photonics for AI and the role of AI for photonics. [Extracted from the article]

In this paper, a data augmentation aided complex-valued network is proposed for underwater acoustic (UWA) orthogonal frequency division multiplexing (OFDM) channel estimations, wherein empirical mode decomposition based data augmentation is proposed to solve the current dilemma in the deep learning embedded UWA-OFDM communications: data scarcity and data-sampling difficulties in real-world applications. In addition, the significance of high-frequency component augmentation for the UWA channel and how it positively influences the following model training are discussed in detail and demonstrated experimentally in this paper. In addition, the complex-valued network is specially designed for the complex-formatted UWA-OFDM signal, which can fully utilize the relationship between its real and imaginary parts with half of the spatial resources of its real-valued counterparts. The experiments with the at-sea-measured WATERMARK dataset indicate that the proposed method can perform a near-optimal channel estimation, and its low resource requirements (on dataset and model) make it more adaptable to real-world UWA applications. [ABSTRACT FROM AUTHOR]

This paper aims at proposing a data-driven Reynolds Averaged Navier–Stokes (RANS) calculation model based on physically constrained deep learning. Using the standard k − ɛ model as the template, part of the source terms in the ɛ equation is replaced by the deep learning model. The simulation results of this new model achieve a high error reduction of 51.7% compared to the standard k − ɛ model. To improve the generality, the accuracy, and the convergence for the undeveloped flow, this paper focuses on optimizing the training process and introducing a data correction method named "coordinate" technology. For the training dataset, the k-field and ɛ-field are automatically corrected by using this technology when the flow state deviates from the theoretical estimation of the standard k − ɛ model. Based on the coordinate technology, a source term of the equation is built by deep learning, and the simulation error is reduced by 6.2% compared to the uncoordinated one. The results confirm that the coordinate technology can effectively adapt to the undeveloped flow where the standard k − ɛ model is not suited and improve the accuracy of the data-driven RANS modeling when dealing with complex flows. [ABSTRACT FROM AUTHOR]

This paper presents a deep learning-based approach for the spatiotemporal reconstruction of sound fields using generative adversarial networks. The method utilises a plane wave basis and learns the underlying statistical distributions of pressure in rooms to accurately reconstruct sound fields from a limited number of measurements. The performance of the method is evaluated using two established datasets and compared to state-of-the-art methods. The results show that the model is able to achieve an improved reconstruction performance in terms of accuracy and energy retention, particularly in the high-frequency range and when extrapolating beyond the measurement region. Furthermore, the proposed method can handle a varying number of measurement positions and configurations without sacrificing performance. The results suggest that this approach provides a promising approach to sound field reconstruction using generative models that allow for a physically informed prior to acoustics problems. [ABSTRACT FROM AUTHOR]

The intelligent data-driven screening of pathological voice signals is a non-invasive and real-time tool for computer-aided diagnosis that has attracted increasing attention from researchers and clinicians. In this paper, the authors propose multi-domain features and the hierarchical extreme learning machine (H-ELM) for the automatic identification of voice disorders. A sufficient number of sensitive features are first extracted from the original voice signal through multi-domain feature extraction (i.e., features of the time domain and the sample entropy based on ensemble empirical mode decomposition and gammatone frequency cepstral coefficients). To eliminate redundancy in high-dimensional features, neighborhood component analysis is then applied to filter out sensitive features from the high-dimensional feature vectors to improve the efficiency of network training and reduce overfitting. The sensitive features thus obtained are then used to train the H-ELM for pathological voice classification. The results of the experiments showed that the sensitivity, specificity, F1 score, and accuracy of the H-ELM were 99.37%, 98.61%, 99.37%, and 98.99%, respectively. Therefore, the proposed method is feasible for the initial classification of pathological voice signals.

Multi-conditional holographic pressure coefficients over a wing are crucial for wing design, and a wind tunnel test is an indispensable means to obtain this profile. However, it is resource-consuming to obtain wind tunnel data under different conditions and only a limited number of sensors can be placed on the wing model during one test, which results in sparse pressure coefficient data with distribution inconsistency across cross sections and conditions. Thus, how to obtain pressure coefficients of more cross sections or even the whole wing surface with multiple conditions from the distribution-inconsistent sensor data becomes a challenging problem. Therefore, a deep learning framework based on transfer learning is proposed in this paper, in which the spatial dependency captured by a long short-term memory model between the obtained multi-conditional sensor data is transferred to other cross sections with few-condition data on the wing. The results demonstrate that the proposed framework achieves high accuracy on the pressure coefficients prediction of distribution-inconsistent cross sections on wind tunnel test data, and thus improves data utilization and cuts costs by reducing wind tunnel tests under different design conditions. Our work proves the possibility of reconstructing the holographic flow field from sparse sensor data of wind tunnel tests and puts forward recommendations on the placement of sensors for achieving this goal. [ABSTRACT FROM AUTHOR]

The radiated noise from ships is of great significance to target recognition, and several deep learning methods have been developed for the recognition of underwater acoustic signals. Previous studies have focused on single-target recognition, with relatively few reports on multitarget recognition. This paper proposes a deep learning-based single-channel multitarget underwater acoustic signal recognition method for an unknown number of targets in the specified category. The proposed method allows the two subproblems of recognizing the unique class and duplicate categories of multiple targets to be solved. These two tasks are essentially multilabel binary classification and multilabel multiple value classification, respectively. In this paper, we describe the use of real-valued and complex-valued ResNet and DenseNet convolutional networks to recognize synthetic mixed multitarget signals, which was superimposed from individual target signals. We compare the performance of various features, including the original audio signal, complex-valued short-time Fourier transform (STFT) spectrum, magnitude STFT spectrum, logarithmic mel spectrum, and mel frequency cepstral coefficients. The experimental results show that our method can effectively recognize synthetic multitarget ship signals when the magnitude STFT spectrum, complex-valued STFT spectrum, and log-mel spectrum are used as network inputs. [ABSTRACT FROM AUTHOR]

For hearing aids, it is critical to reduce the acoustic coupling between the receiver and microphone to ensure that prescribed gains are below the maximum stable gain, thus preventing acoustic feedback. Methods for doing this include fixed and adaptive feedback cancellation, phase modulation, and gain reduction. However, the behavior of hearing aids in situations where the prescribed gain is only just below the maximum stable gain, called here "marginally stable gain," is not well understood. This paper analyzed marginally stable systems and identified three problems, including increased gain at frequencies with the smallest gain margin, short whistles caused by the limited rate of decay of the output when the input drops, and coloration effects. A deep learning framework, called deep marginal feedback cancellation (DeepMFC), was developed to suppress short whistles, and reduce coloration effects, as well as to limit excess amplification at certain frequencies. To implement DeepMFC, many receiver signals in closed-loop systems and corresponding open-loop systems were simulated, and the receiver signals of the closed-loop and open-loop systems were paired together to obtain parallel signals for training. DeepMFC achieved much better performance than existing feedback control methods using objective and subjective measures.

The survivability of autonomous underwater vehicles (AUV) in complex missions and dangerous situations is of great significance to ocean resource exploration, hydrological research, maritime rescue, and undersea military. Existing researches on motion control for the AUV mainly focus on its normal operating, but the active self-rescue method in emergency situations is hardly found. As classical control methods are not sufficient enough for complicated self-rescue missions of the AUV, this paper uses the deep reinforcement learning (DRL) algorithm to solve this problem because the DRL algorithm has the advantages in learning and decision making for complex robot control missions. In this paper, the normal motion control of the AUV based on the deep deterministic policy gradient algorithm is explored, including the yaw angle adjustment, yaw angle adjustment extension, trajectory tracking, and normal floating-up control of the AUV. Then, active self-rescue methods are successfully achieved to recover the AUV from emergencies, such as ocean water density decreasing sharply or one fin getting jammed at a random angle. What is more, real environment experiments are successfully conducted on a self-developed platform of the AUV to validate the feasibility of the proposed control methods. The results can effectively improve the survivability of the AUV and can be a reference to submarine survivability technologies. [ABSTRACT FROM AUTHOR]

The probability distribution of three-dimensional sound speed fields (3D SSFs) in an ocean region encapsulates vital information about their variations, serving as valuable data-driven priors for SSF inversion tasks. However, learning such a distribution is challenging due to the high dimensionality and complexity of 3D SSFs. To tackle this challenge, we propose employing the diffusion model, a cutting-edge deep generative model that has showcased remarkable performance in diverse domains, including image and audio processing. Nonetheless, applying this approach to 3D ocean SSFs encounters two primary hurdles. First, the lack of publicly available well-crafted 3D SSF datasets impedes training and evaluation. Second, 3D SSF data consist of multiple 2D layers with varying variances, which can lead to uneven denoising during the reverse process. To surmount these obstacles, we introduce a novel 3D SSF dataset called 3DSSF, specifically designed for training and evaluating deep generative models. In addition, we devise a high-capacity neural architecture for the diffusion model to effectively handle variations in 3D sound speeds. Furthermore, we employ state-of-the-art continuous-time-based optimization method and predictor-corrector scheme for high-performance training and sampling. Notably, this paper presents the first evaluation of the diffusion model's effectiveness in generating 3D SSF data. Numerical experiments validate the proposed method's strong ability to learn the underlying data distribution of 3D SSFs, and highlight its effectiveness in assisting SSF inversion tasks and subsequently characterizing the transmission loss of underwater acoustics. [ABSTRACT FROM AUTHOR]

Sound field reproduction, which attempts to create a virtual acoustic environment, is a fundamental technology in the achievement of virtual reality. In sound field reproduction, the driving signals of the loudspeakers are calculated by considering the signals collected by the microphones and working environment of the reproduction system. In this paper, an end-to-end reproduction method based on deep learning is proposed. The inputs and outputs of this system are the sound-pressure signals recorded by microphones and the driving signals of loudspeakers, respectively. A convolutional autoencoder network with skip connections in the frequency domain is used. Furthermore, sparse layers are applied to capture the sparse features of the sound field. Simulation results show that the reproduction errors of the proposed method are lower than those generated by the conventional pressure matching and least absolute shrinkage and selection operator methods, especially at high frequencies. Experiments were performed under conditions of single and multiple primary sources. The results in both cases demonstrate that the proposed method achieves better high-frequency performance than the conventional methods. [ABSTRACT FROM AUTHOR]

Masked Autoencoders (MAEs), the state-of-the-art self-supervised neural network architecture in miscellaneous vision tasks, show surprisingly effective potential in reconstructing images distorted by random masking. This paper first introduces an optical implementation of MAEs, employing digital micromirror devices in the optical path to capture partially blocked images. MAEs with multi-scale patches are deployed in the reconstruction procedure. By using an optical-specialized version of the reconstruction network, the system can reconstruct original scenes of high quality. Simulations and experimental measurements showed a significant performance, achieving 24.41 dB average peak-signal-to-noise on Davis2017 datasets and 29.92 dB (masked areas) on authentic captured images under 70% of pixels being blocked. This paves the way for the application of low-bandwidth sampling of high-throughput high-resolution images. [ABSTRACT FROM AUTHOR]

Compressive light field cameras have attracted notable attention over the past few years because they can efficiently determine redundancy from light fields. However, much of the research has only concentrated on reconstructing the entire light field from compressed sampling, which ignores the possibility of directly extracting information such as depth from it. In this paper, we introduce a light field camera configuration with a random color-coded microlens array. Considering the color-coded light fields, we propose a novel attention-based encoder–decoder network. Specifically, the encoder part compresses the coded measurement into a low-dimensional representation that removes most redundancy, and the decoder part constructs the depth map directly from the latent representation. The attention mechanism enables the network to process spatial and angular features dynamically and effectively, thus significantly improving performance. Extensive experiments on synthetic and real-world datasets show that our method outperforms the state-of-the-art light field depth estimation method designed for non-coded light fields. To our knowledge, this is the first study that combines the color-coded light field with the attention-based deep learning approach, which provides a crucial insight into the design of enhanced light field photography systems. [ABSTRACT FROM AUTHOR]

Traditional stereophonic acoustic echo cancellation algorithms need to estimate acoustic echo paths from stereo loudspeakers to a microphone, which often suffers from the nonuniqueness problem caused by a high correlation between the two far-end signals of these stereo loudspeakers. Many decorrelation methods have already been proposed to mitigate this problem. However, these methods may reduce the audio quality and/or stereophonic spatial perception. This paper proposes to use a convolutional recurrent network (CRN) to suppress the stereophonic echo components by estimating a nonlinear gain, which is then multiplied by the complex spectrum of the microphone signal to obtain the estimated near-end speech without a decorrelation procedure. The CRN includes an encoder-decoder module and two-layer gated recurrent network module, which can take advantage of the feature extraction capability of the convolutional neural networks and temporal modeling capability of recurrent neural networks simultaneously. The magnitude spectra of the two far-end signals are used as input features directly without any decorrelation preprocessing and, thus, both the audio quality and stereophonic spatial perception can be maintained. The experimental results in both the simulated and real acoustic environments show that the proposed algorithm outperforms traditional algorithms such as the normalized least-mean square and Wiener algorithms, especially in situations of low signal-to-echo ratio and high reverberation time RT 60 . [ABSTRACT FROM AUTHOR]

When making voice interactions with hands-free speech communication devices, direction-of-arrival estimation is an essential step. To address the detrimental influence of unavoidable background noise and interference speech on direction-of-arrival estimation, this paper introduces a stacked self-attention network system, a supervised deep learning method that enables utterance level estimation without requirement for any pre-processing such as voice activity detection. Specifically, alternately stacked time- and frequency-dependent self-attention blocks are designed to process information in terms of time and frequency, respectively. The former blocks focus on the importance of each time frame of the received audio mixture and perform temporal selection to reduce the influence of non-speech and interference frames, while the latter blocks are utilized to derive inner-correlation among different frequencies. Additionally, the non-causal convolution and self-attention networks are replaced by causal ones, enabling real-time direction-of-arrival estimation with a latency of only 6.25 ms. Experiments with simulated and measured room impulse responses, as well as real recordings, verify the advantages of the proposed method over the state-of-the-art baselines. [ABSTRACT FROM AUTHOR]

Material characterization techniques are widely used to characterize the physical and chemical properties of materials at the nanoscale and, thus, play central roles in material scientific discoveries. However, the large and complex datasets generated by these techniques often require significant human effort to interpret and extract meaningful physicochemical insights. Artificial intelligence (AI) techniques such as machine learning (ML) have the potential to improve the efficiency and accuracy of surface analysis by automating data analysis and interpretation. In this perspective paper, we review the current role of AI in surface analysis and discuss its future potential to accelerate discoveries in surface science, materials science, and interface science. We highlight several applications where AI has already been used to analyze surface analysis data, including the identification of crystal structures from XRD data, analysis of XPS spectra for surface composition, and the interpretation of TEM and SEM images for particle morphology and size. We also discuss the challenges and opportunities associated with the integration of AI into surface analysis workflows. These include the need for large and diverse datasets for training ML models, the importance of feature selection and representation, and the potential for ML to enable new insights and discoveries by identifying patterns and relationships in complex datasets. Most importantly, AI analyzed data must not just find the best mathematical description of the data, but it must find the most physical and chemically meaningful results. In addition, the need for reproducibility in scientific research has become increasingly important in recent years. The advancement of AI, including both conventional and the increasing popular deep learning, is showing promise in addressing those challenges by enabling the execution and verification of scientific progress. By training models on large experimental datasets and providing automated analysis and data interpretation, AI can help to ensure that scientific results are reproducible and reliable. Although integration of knowledge and AI models must be considered for the transparency and interpretability of models, the incorporation of AI into the data collection and processing workflow will significantly enhance the efficiency and accuracy of various surface analysis techniques and deepen our understanding at an accelerated pace. [ABSTRACT FROM AUTHOR]

Conventional microphone array methods for the characterization of sound sources that require a focus-grid are, depending on the grid resolution, either computationally demanding or limited in reconstruction accuracy. This paper presents a deep learning method for grid-free source characterization using a Transformer architecture that is exclusively trained with simulated data. Unlike previous grid-free model architectures, the presented approach requires a single model to characterize an unknown number of ground-truth sources. The model predicts a set of source components, spatially arranged in clusters. Integration over the predicted cluster components allows for the determination of the strength for each ground-truth source individually. Fast and accurate source mapping performance of up to ten sources at different frequencies is demonstrated and strategies to reduce the training effort at neighboring frequencies are given. A comparison with the established grid-based CLEAN-SC and a probabilistic sparse Bayesian learning method on experimental data emphasizes the validity of the approach. [ABSTRACT FROM AUTHOR]

Scanning or transmission electron microscopy (SEM/TEM) are standard techniques used during Research and Development (R&D) phases to study the structure and morphology of microscopic materials. Variety in object shapes and sizes are observed in such images to ensure robust micro- and nanomaterials critical dimension analysis. This way, precision and accuracy can be guaranteed during materials manufacturing processes. Such diversity and complexity in the data make it challenging to automatically extract the desired measurements of these microscopic structures. Existing tools in metrology often require many manual interactions, therefore being slow and prone to user errors. Proposed semiautomatic and automatic tools in the state-of-the-art are also limited and not designed to handle large variations across the images. Thus, the application of advanced machine or deep learning techniques could bring great efficiency in SEM/TEM image analysis and measurements for microscopic scale R&D processes. In this paper, we demonstrate the feasibility of deep-learning based object detection and instance segmentation models to perform automatic and accurate metrology on microscopic images with high object variability. We also show that custom object detection models prepared using pretrained weights, finetuned on very limited custom data, can outperform detection models built using traditional methods. This is particularly useful in metrology for the semiconductor industry, where data scarcity is common. When the data are available, we observe that it can be useful to be able to generate a large number of quality annotations to use instance segmentation. This could allow the training of more complex deep learning models for particle recognition and analysis. Therefore, we propose a semiautomatic tool to help produce annotations and demonstrate its application in an instance segmentation task. [ABSTRACT FROM AUTHOR]

Sound source localization in noisy and reverberant rooms using microphone arrays remains a challenging task, especially for small-sized arrays. Recent years have seen promising advances on deep learning assisted approaches by reformulating the sound localization problem as a classification one. A key to the deep learning-based approaches lies in extracting sound location features effectively in noisy and reverberant conditions. The popularly adopted features are based on the well-established generalized cross correlation phase transform (GCC-PHAT), which is known to be helpful in combating room reverberation. However, the GCC-PHAT features may not be applicable to small-sized arrays. This paper proposes a deep learning assisted sound localization method using a small-sized microphone array constructed by two orthogonal first-order differential microphone arrays. An improved feature extraction scheme based on sound intensity estimation is also proposed by decoupling the correlation between sound pressure and particle velocity components in the whitening weighting construction to enhance the robustness of the time-frequency bin-wise sound intensity features. Simulation and real-world experimental results show that the proposed deep learning assisted approach can achieve higher spatial resolution and is superior to its state-of-the-art counterparts using the GCC-PHAT or sound intensity features for small-sized arrays in noisy and reverberant environments. [ABSTRACT FROM AUTHOR]

This paper investigates the methods to detect and classify marmoset vocalizations automatically using a large data set of marmoset vocalizations and deep learning techniques. For vocalization detection, neural networks-based methods, including deep neural network (DNN) and recurrent neural network with long short-term memory units, are designed and compared against a conventional rule-based detection method. For vocalization classification, three different classification algorithms are compared, including a support vector machine (SVM), DNN, and long short-term memory recurrent neural networks (LSTM-RNNs). A 1500-min audio data set containing recordings from four pairs of marmoset twins and manual annotations is employed for experiments. Two test sets are built according to whether the test samples are produced by the marmosets in the training set (test set I) or not (test set II). Experimental results show that the LSTM-RNN-based detection method outperformed others and achieved 0.92% and 1.67% frame error rate on these two test sets. Furthermore, the deep learning models obtained higher classification accuracy than the SVM model, which was 95.60% and 91.67% on the two test sets, respectively. [ABSTRACT FROM AUTHOR]

Realistic wind data are essential in developing, testing, and ensuring the safety of unmanned aerial systems in operation. Alternatives to Dryden and von Kármán turbulence models are required, aimed explicitly at urban air spaces to generate turbulent wind data. We present a novel method to generate realistic wind data for the safe operation of small unmanned aerial vehicles in urban spaces. We propose a non-intrusive reduced order modeling approach to replicate realistic wind data and predict wind fields. The method uses a well-established large-eddy simulation model, the parallelized large eddy simulation model, to generate high-fidelity data. To create a reduced-order model, we utilize proper orthogonal decomposition to extract modes from the three-dimensional space and use specialized recurrent neural networks and long-term short memory for stepping in time. This paper combines the traditional approach of using computational fluid dynamic simulations to generate wind data with deep learning and reduced-order modeling techniques to devise a methodology for a non-intrusive data-based model for wind field prediction. A simplistic model of an isolated urban subspace with a single building setup in neutral atmospheric conditions is considered a test case for the demonstration of the method. [ABSTRACT FROM AUTHOR]

This article is a survey of deep learning methods for single and multiple sound source localization, with a focus on sound source localization in indoor environments, where reverberation and diffuse noise are present. We provide an extensive topography of the neural network-based sound source localization literature in this context, organized according to the neural network architecture, the type of input features, the output strategy (classification or regression), the types of data used for model training and evaluation, and the model training strategy. Tables summarizing the literature survey are provided at the end of the paper, allowing a quick search of methods with a given set of target characteristics. [ABSTRACT FROM AUTHOR]

Understanding the dynamic system that produces speech is essential to advancing speech science, and several simultaneous sensory streams can be leveraged to describe the process. As the tongue functional deformation correlates with the lip's shapes of the speaker, this paper aims to explore the association between them. The problem is formulated as a sequence to sequence learning task and a deep neural network is trained using unlabeled lip videos to predict an upcoming ultrasound tongue image sequence. Experimental results show that the machine learning model can predict the tongue's motion with satisfactory performance, which demonstrates that the learned neural network can build the association between two imaging modalities. [ABSTRACT FROM AUTHOR]

During the past decade, deep learning has represented the biggest research trend in the field of machine learning, which provides new powerful tools to interrogate high dimensional time series data in a way that has not been possible before. With the recent success in natural language processing, one would expect widespread adaptation to problems like time series forecasting and classification. After all, both involve processing sequential data. However, to this point, research on their adaptation to time series problems has remained limited. Recently, a multi-step time-stepping scheme without the need of direct access to temporal gradients has been proposed, which can accurately identify nonlinear dynamical systems from time series data. In this paper, we combined an attention mechanism with a deep model in a multi-step time-stepping scheme to perform nonlinear system identification and forecasting tasks. To our knowledge, this is the first paper to use attention based models to deal with nonlinear system identification and forecasting problems. The attention weights on rows select those variables that are helpful for forecasting, which can enhance the information across multiple time steps to capture temporal information. The experiment results indicate that the attention based model in a multi-step time-stepping scheme has better identification and prediction performance for nonlinear time series identification and forecasting problems. [ABSTRACT FROM AUTHOR]

This paper presents a deep-learning-based algorithm dedicated to the processing of speckle noise in phase measurements in digital holographic interferometry. The deep learning architecture is trained with phase fringe patterns including faithful speckle noise, having non-Gaussian statistics and non-stationary property, and exhibiting spatial correlation length. The performances of the speckle de-noiser are estimated with metrics, and the proposed approach exhibits state-of-the-art results. In order to train the network to de-noise phase fringe patterns, a database is constituted with a set of noise-free and speckled phase data. The algorithm is applied to de-noising experimental data from wide-field digital holographic vibrometry. Comparison with the state-of-the-art algorithm confirms the achieved performance. [ABSTRACT FROM AUTHOR]

This paper proposes a robust system for detecting North Atlantic right whales by using deep learning methods to denoise noisy recordings. Passive acoustic recordings of right whale vocalisations are subject to noise contamination from many sources, such as shipping and offshore activities. When such data are applied to uncompensated classifiers, accuracy falls substantially. To build robustness into the detection process, two separate approaches that have proved successful for image denoising are considered. Specifically, a denoising convolutional neural network and a denoising autoencoder, each of which is applied to spectrogram representations of the noisy audio signal, are developed. Performance is improved further by matching the classifier training to include the vestigial signal that remains in clean estimates after the denoising process. Evaluations are performed first by adding white, tanker, trawler, and shot noises at signal-to-noise ratios from -10 to +5 dB to clean recordings to simulate noisy conditions. Experiments show that denoising gives substantial improvements to accuracy, particularly when using the vestigial-trained classifier. A final test applies the proposed methods to previously unseen noisy right whale recordings and finds that denoising is able to improve performance over the baseline clean-trained model in this new noise environment.

Guided ultrasonic wave localization systems use spatially distributed sensor arrays and wave propagation models to detect and locate damage across a structure. Environmental and operational conditions, such as temperature or stress variations, introduce uncertainty into guided wave data and reduce the effectiveness of these localization systems. These uncertainties cause the models used by each localization algorithm to fail to match with reality. This paper addresses this challenge with an ensemble deep neural network that is trained solely with simulated data. Relative to delay-and-sum and matched field processing strategies, this approach is demonstrated to be more robust to temperature variations in experimental data. As a result, this approach demonstrates superior accuracy with small numbers of sensors and greater resilience to spatially nonhomogeneous temperature variations over time. [ABSTRACT FROM AUTHOR]

This review presents a high-level overview of the uses of machine learning (ML) to address several challenges in spatial auditory display research, primarily using head-related transfer functions. This survey also reviews and compares several categories of ML techniques and their application to virtual auditory reality research. This work addresses the use of ML techniques such as dimensionality reduction, unsupervised learning, supervised learning, reinforcement learning, and deep learning algorithms. The paper concludes with a discussion of the usage of ML algorithms to address specific spatial auditory display research challenges. [ABSTRACT FROM AUTHOR]

Deep learning-based fringe projection profilometry (FPP) shows potential for challenging three-dimensional (3D) reconstruction of objects with dynamic motion, complex surface, and extreme environment. However, the previous deep learning-based methods are all supervised ones, which are difficult to be applied for scenes that are different from the training, thus requiring a large number of training datasets. In this paper, we propose a new geometric constraint-based phase unwrapping (GCPU) method that enables an untrained deep learning-based FPP for the first time. An untrained convolutional neural network is designed to achieve correct phase unwrapping through a network parameter space optimization. The loss function of the optimization is constructed by following the 3D, structural, and phase consistency. The designed untrained network directly outputs the desired fringe order with the inputted phase and fringe background. The experiments verify that the proposed GCPU method provides higher robustness compared with the traditional GCPU methods, thus resulting in accurate 3D reconstruction for objects with a complex surface. Unlike the commonly used temporal phase unwrapping, the proposed GCPU method does not require additional fringe patterns, which can also be used for the dynamic 3D measurement. [ABSTRACT FROM AUTHOR]

In this paper, we develop a deep learning approach for the accurate solution of challenging problems of near-field microscopy that leverages the powerful framework of physics-informed neural networks (PINNs) for the inversion of the complex optical parameters of nanostructured environments. Specifically, we show that PINNs can be flexibly designed based on full-vector Maxwell's equations to inversely retrieve the spatial distributions of the complex electric permittivity and magnetic permeability of unknown scattering objects in the resonance regime from near-field data. Moreover, we demonstrate that PINNs achieve excellent convergence to the true material parameters under both plane wave and point source (localized) excitations, enabling parameter retrieval in scanning near-field optical microscopy. Our method is computationally efficient compared to traditional data-driven deep learning approaches as it requires only a single dataset for training. Furthermore, we develop and successfully demonstrate adaptive PINNs with trainable loss weights that largely improve the accuracy of the inverse reconstruction for high-index materials compared to standard PINNs. Finally, we demonstrate the full potential of our approach by retrieving the space-dependent permittivity of a three-dimensional unknown object from near-field data. The presented framework paves the way to the development of a computationally driven, accurate, and non-invasive platform for the simultaneous retrieval of the electric and magnetic parameters of resonant nanostructures from measured optical images, with applications to biomedical imaging, optical remote sensing, and characterization of metamaterial devices. [ABSTRACT FROM AUTHOR]

Ultrasonic non-destructive testing can effectively detect damage in aircraft composite materials, but traditional manual testing is time-consuming and labor-intensive. To realize the intelligent recognition of aircraft composite material damage, this paper proposes a 1D-YOLO network, in which intelligent fusion recognizes both the ultrasonic C-scan image and ultrasonic A-scan signal of composite material damage. Through training and testing the composite material damage data on aircraft skin, the accuracy of the model is 94.5%, the mean average precision is 80.0%, and the kappa value is 97.5%. The use of dilated convolution and a recursive feature pyramid effectively improves the feature extraction ability of the model. The effectively used Cascade R-CNN (Cascade Region-Convolutional Neural Network) improves the recognition effect of the model, and the effectively used one-dimensional convolutional neural network excludes non-damaged objects. Comparing our network with YOLOv3, YOLOv4, cascade R-CNN, and other networks, the results show that our network can identify the damage of composite materials more accurately. [ABSTRACT FROM AUTHOR]

This paper is part of a special issue on machine learning in acoustics. A model-based convolutional neural network (CNN) approach is presented to test the viability of this method as an alternative to conventional matched-field processing (MFP) for underwater source-range estimation. The networks are trained with simulated data generated under a particular model of the environment. When tested with data simulated in environments that deviate slightly from the training environment, this approach shows improved prediction accuracy and lower mean-absolute-error (MAE) compared to MFP. The performance of this model-based approach also transfers to real data, as demonstrated separately with field data collected in the Beaufort Sea and off the coast of Southern California. For the former, the CNN predictions are consistent with expected source range while for the latter, the CNN estimates have lower MAE compared to MFP. Examination of the trained CNNs' intermediate outputs suggests that the approach is more constrained than MFP from outputting very inaccurate predictions when there is a slight environmental mismatch. This improvement appears to be at the expense of decreased certainty in the correct source range prediction when the environment is precisely modeled.

The aircraft skin is an important component of the aircraft, and its integrity affects the flight performance and safety performance of the aircraft. Damage detection technology with ultrasonic nondestructive testing as the core has played an important role in aircraft skin damage detection. Due to the difficulty in aircraft skin detection, relying solely on the ultrasonic A-scan equipment has very low detection efficiency. The introduction of artificial intelligence can effectively improve the detection efficiency. This paper establishes the one-dimensional convolutional neural network and single shot multibox detector network, which is based on the SSD network and uses dilated convolution to improve the real-time tracking of the ultrasonic probe. At the same time, 1DCNN is introduced to classify the ultrasonic A-scan signal. Finally, the detection results of the two are combined to achieve the reflection of the internal conditions of the aircraft skin. After testing, the algorithm can identify skin simulation specimens, and its recognition accuracy is 96.5%, AP is 90.9%, and kappa is 0.996. Comparing the improved SSD network with networks such as SSD, YOLOv3, and Faster R-CNN, the results show that the improved network used in this paper is more excellent and effective. At the same time, the detection effects of four types of optimization algorithms and five learning rates are studied, and finally, the corresponding signal classification model adopts Adam and the learning rate of 0.0001 has the best effect. [ABSTRACT FROM AUTHOR]

In this paper, we introduce an extension of the image method for generating room impulse responses in a structure with more than a single confined space, namely, the structure image method (StIM). The proposed method, StIM, can efficiently generate a large number of environmental examples for a structure impulse response, which is required by current deep-learning methods for many tasks, while maintaining low computational complexity. We address the integration of the environment representation, produced by StIM, into the training process, and present a framework for training deep models. We demonstrate the usage of StIM when training an audio classification model and testing with real recordings acquired by accessible day-to-day devices. StIM shows promising results for indoors audio classification, where the target sound source is not located in the same room as the microphones. StIM enables large scale simulations of multi-room acoustics with low computational complexity which is mostly beneficial for training of deep learning networks. [ABSTRACT FROM AUTHOR]