Search

Yixuan Zhang,1,* Min Wang,2,* Ting Zhang,3 Honghua Wang,2 Ying Chen,1 Tao Zhou,1 Rui Yang1 1Research Institute for Reproductive Medicine and Genetic Diseases, Wuxi Maternity and Child Health Care Hospital, Wuxi, 214002, People’s Republic of China; 2Center for Reproductive Medicine, The Affiliated Wuxi Maternity and Child Health Care Hospital of Nanjing Medical University, Wuxi, 214002, People’s Republic of China; 3Department of Laboratory Medicine, Wuxi Maternity and Child Health Care Hospital, Jiangnan University, Wuxi, 214002, People’s Republic of China*These authors contributed equally to this workCorrespondence: Tao Zhou; Rui Yang, Email zhoutao@njmu.edu.cn; ruiyang826@163.comAbstract: Sperm quality is declining dramatically during the past decades. Male infertility has been a serious health and social problem. The sperm cell driven biohybrid nanorobot opens a new era for automated and precise assisted reproduction. Therefore, it is urgent and necessary to conduct an updated review and perspective from the viewpoints of the researchers and clinicians in the field of reproductive medicine. In the present review, we first update the current classification, design, control and applications of various spermbots. Then, by a comprehensive summary of the functional features of sperm cells, the journey of sperms to the oocyte, and sperm-related dysfunctions, we provide a systematic guidance to further improve the design of spermbots. Focusing on the translation of spermbots into clinical practice, we point out that the main challenges are biocompatibility, effectiveness, and ethical issues. Considering the special requirements of assisted reproduction, we also propose the three laws for the clinical usage of spermbots: good genetics, gentle operation and no contamination. Finally, a three-step roadmap is proposed to achieve the goal of clinical translation. We believe that spermbot-based treatments can be validated and approved for in vitro

During the artificial breeding process of Chinese pearl oysters, the quality and vitality of sperm and eggs are pivotal in determining the success rate of fertilization. This study delves into the impact of externally added fructose, glucose, and ammoniated water on sperm vitality and lifespan under varying temperature conditions. The results indicate that the addition of exogenous energy sources and cell protectants, specifically monosaccharide energy substances, significantly improves sperm vitality. Moreover, under 4°C conditions, adding 2.8 mmol/L fructose, glucose, and 2 mmol/L ammoniated water to sperm samples significantly enhances sperm vitality and prolongs their lifespan. This provides strong evidence for the feasibility of short-term preservation and vitality enhancement strategies for Chinese pearl oyster sperm and also offers important scientific support for optimizing the artificial reproduction techniques of Chinese pearl oysters.

Co-N co-doped materials have been widely used in PMS systems for organic pollutant degradation. However, the synergy effect of Co and N in PMS activation processes is not fully understood. In the present study, a catalyst comprising Co nanoparticles incorporated within N-doped biochar (Co@NBC) was synthesized to facilitate the activation of PMS and the degradation of norfloxacin (NOR). The results demonstrated that the degradation efficiency of NOR in the Co@NBC-PMS system was 94.45% (0.341 min-1). Quenching tests, EPR analyses, electrochemical measurements, and DFT calculations indicated that large amounts of graphitic N in Co@NBC remarkably augmented the direct electron transfer (DET) process. Furthermore, oxidized graphitic N stretched the O-H bond of PMS and promoted the generation of SO5(center dot)-, which was the main intermediate product for the formation of 1O2. In addition, graphitic N could facilitate the conversion of CoII to CoIV 00000000 00000000 00000000 00000000 11111111 00000000 11111111 00000000 00000000 00000000 O by elongating the S-O bond of PMS and shortening the Co-O bond. Besides, reduced pyridinic N encouraged the reduction of CoIVO to CoIII and CoII. This valence state cycle of CoII-CoIVO-CoIII co-promoted by graphitic N and pyridinic N significantly enhanced the PMS activation process. Moreover, the possible degradation pathways of NOR and the environmental impact of its intermediates had been investigated through DFT calculations, LC-MS, and ECOSAR. Importantly, the developed system exhibited efficient removal of NOR under inorganic ion influence and in practical water matrices. This study not only developed an efficient PMS activation system (Co@NBC-PMS) for the remediation of antibiotic-contaminated wastewater, but also offered novel insights into the synergy effect of Co and N on activation mechanisms. Co-N co-doped biochar synergistically activated PMS for high-efficiency degradation of NOR. Graphitic N and pyridinic N co-promoted the

Osteosarcoma(OS) is a highly aggressive bone cancer for which treatment has remained essentially unchanged for decades. Although OS is characterized by extensive genomic heterogeneity and instability, RB1 and TP53 have been shown to be the most commonly inactivated tumor suppressors in OS. We previously generated a mouse model with a double knockout (DKO) of Rb1 and Trp53 within cells of the osteoblastic lineage, which largely recapitulates human OS with nearly complete penetrance. SKP2 is a repression target of pRb and serves as a substrate recruiting subunit of the SCFSKP2 complex. In addition, SKP2 plays a central role in regulating the cell cycle by ubiquitinating and promoting the degradation of p27. We previously reported the DKOAA transgenic model, which harbored a knock-in mutation in p27 that impaired its binding to SKP2. Here, we generated a novel p53-Rb1-SKP2 triple-knockout model (TKO) to examine SKP2 function and its potential as a therapeutic target in OS. First, we observed that OS tumorigenesis was significantly delayed in TKO mice and their overall survival was markedly improved. In addition, the loss of SKP2 also promoted an apoptotic microenvironment and reduced the stemness of DKO tumors. Furthermore, we found that small-molecule inhibitors of SKP2 exhibited anti-tumor activities in vivo and in OS organoids as well as synergistic effects when combined with a standard chemotherapeutic agent. Taken together, our results suggest that SKP2 inhibitors may reduce the stemness plasticity of OS and should be leveraged as next-generation adjuvants in this cancer.

A method for the approximate merging of disk Wang-Ball (DWB) curves based on the modified snake optimizer (BEESO) is proposed in this paper to address the problem of difficulties in the merging of DWB curves. By extending the approximate merging problem for traditional curves to disk curves and viewing it as an optimization problem, an approximate merging model is established to minimize the merging error through an error formulation. Considering the complexity of the model built, a BEESO with better convergence accuracy and convergence speed is introduced, which combines the snake optimizer (SO) and three strategies including bi-directional search, evolutionary population dynamics, and elite opposition-based learning. The merging results and merging errors of numerical examples demonstrate that BEESO is effective in solving approximate merging models, and it provides a new method for the compression and transfer of product shape data in Computer-Aided Geometric Design., Funding Agencies|National Natural Science Foundation of China

Debiao Xiang,1– 3 Lili Zhou,4 Rui Yang,1,4 Fang Yuan,1– 3 Yilin Xu,4 Yuan Yang,1,4 Yong Qiao,1– 3 Xin Li1– 3 1Department of Pharmacy, The Third Hospital of Changsha, Changsha, Hunan Province, People’s Republic of China; 2Hunan Provincial Key Laboratory of Anti-Resistance Microbial Drugs, Changsha, Hunan Province, People’s Republic of China; 3The Clinical Application Research Institute of Antibiotics in Changsha, Changsha, Hunan Province, People’s Republic of China; 4College of Pharmacy, Hunan University of Chinese Medicine, Changsha, Hunan Province, People’s Republic of ChinaCorrespondence: Xin Li, 176 Laodong West Road, Tianxin District, Changsha, Hunan Province, People’s Republic of China, Tel +86 0731-85171320, Email xin-li@cssdsyy.comAbstract: Currently, cancer remains one of the most significant threats to human health. Treatment of most cancers remains challenging, despite the implementation of diverse therapies in clinical practice. In recent years, research on the mechanism of ferroptosis has presented novel perspectives for cancer treatment. Ferroptosis is a regulated cell death process caused by lipid peroxidation of membrane unsaturated fatty acids catalyzed by iron ions. The rapid development of bio-nanotechnology has generated considerable interest in exploiting iron-induced cell death as a new therapeutic target against cancer. This article provides a comprehensive overview of recent advancements at the intersection of iron-induced cell death and bionanotechnology. In this respect, the mechanism of iron-induced cell death and its relation to cancer are summarized. Furthermore, the feasibility of a nano-drug delivery system based on iron-induced cell death for cancer treatment is introduced and analyzed. Secondly, strategies for inducing iron-induced cell death using nanodrug delivery technology are discussed, including promoting Fenton reactions, inhibiting glutathione peroxidase 4, reducing low glutathione levels, and inhibiting syst

Ferroelectric tunnel junctions are promising towards high-reliability and low-power non-volatile memories and computing devices. Yet it is challenging to maintain a high tunnelling electroresistance when the ferroelectric layer is thinned down towards atomic scale because of the ferroelectric structural instability and large depolarization field. Here we report ferroelectric tunnel junctions based on samarium-substituted layered bismuth oxide, which can maintain tunnelling electroresistance of 7 × 105 with the samarium-substituted bismuth oxide film down to one nanometer, three orders of magnitude higher than previous reports with such thickness, owing to efficient barrier modulation by the large ferroelectric polarization. These ferroelectric tunnel junctions demonstrate up to 32 resistance states without any write-verify technique, high endurance (over 5 × 109), high linearity of conductance modulation, and long retention time (10 years). Furthermore, tunnelling electroresistance over 109 is achieved in ferroelectric tunnel junctions with 4.6-nanometer samarium-substituted bismuth oxide layer, which is higher than commercial flash memories. The results show high potential towards multi-level and reliable non-volatile memories.

We report the detection of gamma-ray emission by the Fermi Large Area Telescope (Fermi-LAT) towards the young massive star cluster RCW 38 in the 1-500 GeV photon energy range. We found spatially extended GeV emission towards the direction of RCW 38, which is best modelled by a Gaussian disc of 0.23$\deg$ radius with a significance of the extension is $\sim 11.4 \sigma$. Furthermore, the spatial correlation with the ionized and molecular gas content favors the hadronic origin of the gamma-ray emission. The gamma-ray spectrum of RCW 38 has a relatively hard photon index of $2.44 \pm 0.03$, which is similar to other young massive star clusters. We argue that the diffuse GeV gamma-ray emission in this region likely originates from the interaction of accelerated protons in the stellar cluster with the ambient gas., Comment: 8 pages, 5 figures, accepted for publication in MNRAS. arXiv admin note: text overlap with arXiv:2210.01352

Quantum machine learning has the potential to provide powerful algorithms for artificial intelligence. The pursuit of quantum advantage in quantum machine learning is an active area of research. For current noisy intermediate-scale quantum computers, various quantum-classical hybrid algorithms have been proposed. One such previously proposed hybrid algorithm is a gate-based variational embedding classifier, which is composed of a classical neural network and a parameterized gate-based quantum circuit. We propose a quantum variational embedding classifier based on an analog quantum computer, where control signals vary continuously in time: our particular focus is an implementation using quantum annealers. In our algorithm, the classical data are transformed into the parameters of the time-varying Hamiltonian of the analog quantum computer by a linear transformation. The nonlinearity needed for a nonlinear classification problem is purely provided by the analog quantum computer through the nonlinear dependence of the final quantum state on the control parameters of the Hamiltonian. We perform numerical simulations that demonstrate the effectiveness of our algorithm for performing binary and multiclass classification on linearly inseparable datasets such as concentric circles and MNIST digits. Our classifier can reach accuracy comparable with that of the best classical classifiers. We find that the performance of our classifier can be increased by increasing the number of qubits, until the performance saturates and fluctuates. Moreover, the number of optimization parameters of our classifier scales linearly with the number of qubits. The increase of the number of training parameters when the size of our model increases is therefore not as fast as that of a neural network. Our algorithm presents the possibility of using current quantum annealers for solving practical machine-learning problems, and it could also be useful to explore quantum advantage in quantum machine l

Yunxiang Hu,1,2,* Rui Yang,1,2,* Sanmao Liu,1,2,* Zefeng Song,3 Hong Wang1,2 1Department of Orthopedics, Central Hospital of Dalian University of Technology, Dalian City, Liaoning Province, People’s Republic of China; 2School of Graduates, Dalian Medical University, Dalian City, Liaoning Province, People’s Republic of China; 3School of Graduates, Dalian University of Technology, Dalian City, Liaoning Province, People’s Republic of China*These authors contributed equally to this workCorrespondence: Hong WangDepartment of Orthopedics, Central Hospital of Dalian University of Technology, No. 826, Southwestern Road, Shahekou District, Dalian City, Liaoning Province, People’s Republic of China, Email wanghong@dmu.edu.cnAbstract: Low back pain (LBP) is a common condition that has substantial consequences on individuals and society, both socially and economically. The primary contributor to LBP is often identified as intervertebral disc degeneration (IVDD), which worsens and leads to significant spinal problems. The conventional treatment approach for IVDD involves physiotherapy, drug therapy for pain management, and, in severe cases, surgery. However, none of these treatments address the underlying cause of the condition, meaning that they cannot fundamentally reverse IVDD or restore the mechanical function of the spine. Nanotechnology and regenerative medicine have made significant advancements in the field of healthcare, particularly in the area of nanodrug delivery systems (NDDSs). These approaches have demonstrated significant potential in enhancing the efficacy of IVDD treatments by providing benefits such as high biocompatibility, biodegradability, precise drug delivery to targeted areas, prolonged drug release, and improved therapeutic results. The advancements in different NDDSs designed for delivering various genes, cells, proteins and therapeutic drugs have opened up new opportunities for effectively addressing IVDD. This

Learned Index Structures (LIS) view a sorted index as a model that learns the data distribution, takes a data element key as input, and outputs the predicted position of the key. The original LIS can only handle lookup operations with no support for updates, rendering it impractical to use for typical workloads. To address this limitation, recent studies have focused on designing efficient dynamic learned indexes. ALEX, as the pioneering dynamic learned index structures, enables dynamism by incorporating a series of design choices, including adaptive key space partitioning, dynamic model retraining, and sophisticated engineering and policies that prioritize read/write performance. While these design choices offer improved average-case performance, the emphasis on flexibility and performance increases the attack surface by allowing adversarial behaviors that maximize ALEX's memory space and time complexity in worst-case scenarios. In this work, we present the first systematic investigation of algorithmic complexity attacks (ACAs) targeting the worst-case scenarios of ALEX. We introduce new ACAs that fall into two categories, space ACAs and time ACAs, which target the memory space and time complexity, respectively. First, our space ACA on data nodes exploits ALEX's gapped array layout and uses Multiple-Choice Knapsack (MCK) to generate an optimal adversarial insertion plan for maximizing the memory consumption at the data node level. Second, our space ACA on internal nodes exploits ALEX's catastrophic cost mitigation mechanism, causing an out-of-memory error with only a few hundred adversarial insertions. Third, our time ACA generates pathological insertions to increase the disparity between the actual key distribution and the linear models of data nodes, deteriorating the runtime performance by up to 1,641X compared to ALEX operating under legitimate workloads., Comment: VLDB 2024

Offline Goal-Conditioned RL (GCRL) offers a feasible paradigm for learning general-purpose policies from diverse and multi-task offline datasets. Despite notable recent progress, the predominant offline GCRL methods, mainly model-free, face constraints in handling limited data and generalizing to unseen goals. In this work, we propose Goal-conditioned Offline Planning (GOPlan), a novel model-based framework that contains two key phases: (1) pretraining a prior policy capable of capturing multi-modal action distribution within the multi-goal dataset; (2) employing the reanalysis method with planning to generate imagined trajectories for funetuning policies. Specifically, we base the prior policy on an advantageweighted conditioned generative adversarial network, which facilitates distinct mode separation, mitigating the pitfalls of out-of-distribution (OOD) actions. For further policy optimization, the reanalysis method generates high-quality imaginary data by planning with learned models for both intra-trajectory and inter-trajectory goals. With thorough experimental evaluations, we demonstrate that GOPlan achieves state-of-the-art performance on various offline multi-goal navigation and manipulation tasks. Moreover, our results highlight the superior ability of GOPlan to handle small data budgets and generalize to OOD goals.

Large Language Models (LLMs) have significantly advanced healthcare innovation on generation capabilities. However, their application in real clinical settings is challenging due to potential deviations from medical facts and inherent biases. In this work, we develop an augmented LLM framework, KG-Rank, which leverages a medical knowledge graph (KG) with ranking and re-ranking techniques, aiming to improve free-text question-answering (QA) in the medical domain. Specifically, upon receiving a question, we initially retrieve triplets from a medical KG to gather factual information. Subsequently, we innovatively apply ranking methods to refine the ordering of these triplets, aiming to yield more precise answers. To the best of our knowledge, KG-Rank is the first application of ranking models combined with KG in medical QA specifically for generating long answers. Evaluation of four selected medical QA datasets shows that KG-Rank achieves an improvement of over 18% in the ROUGE-L score. Moreover, we extend KG-Rank to open domains, where it realizes a 14% improvement in ROUGE-L, showing the effectiveness and potential of KG-Rank.

Abrupt maneuvers by surrounding vehicles (SVs) can typically lead to safety concerns and affect the task efficiency of the ego vehicle (EV), especially with model uncertainties stemming from environmental disturbances. This paper presents a real-time fail-operational controller that ensures the asymptotic convergence of an uncertain EV to a safe state, while preserving task efficiency in dynamic environments. An incremental Bayesian learning approach is developed to facilitate online learning and inference of changing environmental disturbances. Leveraging disturbance quantification and constraint transformation, we develop a stochastic fail-operational barrier based on the control barrier function (CBF). With this development, the uncertain EV is able to converge asymptotically from an unsafe state to a defined safe state with probabilistic stability. Subsequently, the stochastic fail-operational barrier is integrated into an efficient fail-operational controller based on quadratic programming (QP). This controller is tailored for the EV operating under control constraints in the presence of environmental disturbances, with both safety and efficiency objectives taken into consideration. We validate the proposed framework in connected cruise control (CCC) tasks, where SVs perform aggressive driving maneuvers. The simulation results demonstrate that our method empowers the EV to swiftly return to a safe state while upholding task efficiency in real time, even under time-varying environmental disturbances., Comment: 8 pages, 8 figures, accepted for publication in the 22nd European Control Conference (ECC 2024)

Precision in identifying and differentiating micro and macro blood vessels in the retina is crucial for the diagnosis of retinal diseases, although it poses a significant challenge. Current autoencoding-based segmentation approaches encounter limitations as they are constrained by the encoder and undergo a reduction in resolution during the encoding stage. The inability to recover lost information in the decoding phase further impedes these approaches. Consequently, their capacity to extract the retinal microvascular structure is restricted. To address this issue, we introduce Swin-Res-Net, a specialized module designed to enhance the precision of retinal vessel segmentation. Swin-Res-Net utilizes the Swin transformer which uses shifted windows with displacement for partitioning, to reduce network complexity and accelerate model convergence. Additionally, the model incorporates interactive fusion with a functional module in the Res2Net architecture. The Res2Net leverages multi-scale techniques to enlarge the receptive field of the convolutional kernel, enabling the extraction of additional semantic information from the image. This combination creates a new module that enhances the localization and separation of micro vessels in the retina. To improve the efficiency of processing vascular information, we've added a module to eliminate redundant information between the encoding and decoding steps. Our proposed architecture produces outstanding results, either meeting or surpassing those of other published models. The AUC reflects significant enhancements, achieving values of 0.9956, 0.9931, and 0.9946 in pixel-wise segmentation of retinal vessels across three widely utilized datasets: CHASE-DB1, DRIVE, and STARE, respectively. Moreover, Swin-Res-Net outperforms alternative architectures, demonstrating superior performance in both IOU and F1 measure metrics.

In the domain of Natural Language Processing (NLP), Large Language Models (LLMs) have demonstrated promise in text-generation tasks. However, their educational applications, particularly for domain-specific queries, remain underexplored. This study investigates LLMs' capabilities in educational scenarios, focusing on concept graph recovery and question-answering (QA). We assess LLMs' zero-shot performance in creating domain-specific concept graphs and introduce TutorQA, a new expert-verified NLP-focused benchmark for scientific graph reasoning and QA. TutorQA consists of five tasks with 500 QA pairs. To tackle TutorQA queries, we present CGLLM, a pipeline integrating concept graphs with LLMs for answering diverse questions. Our results indicate that LLMs' zero-shot concept graph recovery is competitive with supervised methods, showing an average 3% F1 score improvement. In TutorQA tasks, LLMs achieve up to 26% F1 score enhancement. Moreover, human evaluation and analysis show that CGLLM generates answers with more fine-grained concepts.

Enforcing safety while preventing overly conservative behaviors is essential for autonomous vehicles to achieve high task performance. In this paper, we propose a barrier-enhanced homotopic parallel trajectory optimization (BHPTO) approach with over-relaxed alternating direction method of multipliers (ADMM) for real-time integrated decision-making and planning. To facilitate safety interactions between the ego vehicle (EV) and surrounding vehicles, a spatiotemporal safety module exhibiting bi-convexity is developed on the basis of barrier function. Varying barrier coefficients are adopted for different time steps in a planning horizon to account for the motion uncertainties of surrounding HVs and mitigate conservative behaviors. Additionally, we exploit the discrete characteristics of driving maneuvers to initialize nominal behavior-oriented free-end homotopic trajectories based on reachability analysis, and each trajectory is locally constrained to a specific driving maneuver while sharing the same task objectives. By leveraging the bi-convexity of the safety module and the kinematics of the EV, we formulate the BHPTO as a bi-convex optimization problem. Then constraint transcription and over-relaxed ADMM are employed to streamline the optimization process, such that multiple trajectories are generated in real time with feasibility guarantees. Through a series of experiments, the proposed development demonstrates improved task accuracy, stability, and consistency in various traffic scenarios using synthetic and real-world traffic datasets.

With the development of AI-Generated Content (AIGC), text-to-audio models are gaining widespread attention. However, it is challenging for these models to generate audio aligned with human preference due to the inherent information density of natural language and limited model understanding ability. To alleviate this issue, we formulate the BATON, a framework designed to enhance the alignment between generated audio and text prompt using human preference feedback. Our BATON comprises three key stages: Firstly, we curated a dataset containing both prompts and the corresponding generated audio, which was then annotated based on human feedback. Secondly, we introduced a reward model using the constructed dataset, which can mimic human preference by assigning rewards to input text-audio pairs. Finally, we employed the reward model to fine-tune an off-the-shelf text-to-audio model. The experiment results demonstrate that our BATON can significantly improve the generation quality of the original text-to-audio models, concerning audio integrity, temporal relationship, and alignment with human preference.

Background: Cognitive biases in clinical decision-making significantly contribute to errors in diagnosis and suboptimal patient outcomes. Addressing these biases presents a formidable challenge in the medical field. Objective: This study explores the role of large language models (LLMs) in mitigating these biases through the utilization of a multi-agent framework. We simulate the clinical decision-making processes through multi-agent conversation and evaluate its efficacy in improving diagnostic accuracy. Methods: A total of 16 published and unpublished case reports where cognitive biases have resulted in misdiagnoses were identified from the literature. In the multi-agent framework, we leveraged GPT-4 to facilitate interactions among four simulated agents to replicate clinical team dynamics. Each agent has a distinct role: 1) To make the final diagnosis after considering the discussions, 2) The devil's advocate and correct confirmation and anchoring bias, 3) The tutor and facilitator of the discussion to reduce premature closure bias, and 4) To record and summarize the findings. A total of 80 simulations were evaluated for the accuracy of initial diagnosis, top differential diagnosis and final two differential diagnoses. Results: In a total of 80 responses evaluating both initial and final diagnoses, the initial diagnosis had an accuracy of 0% (0/80), but following multi-agent discussions, the accuracy for the top differential diagnosis increased to 71.3% (57/80), and for the final two differential diagnoses, to 80.0% (64/80). Conclusions: The framework demonstrated an ability to re-evaluate and correct misconceptions, even in scenarios with misleading initial investigations. The LLM-driven multi-agent conversation framework shows promise in enhancing diagnostic accuracy in diagnostically challenging medical scenarios., Comment: 21 pages, 3 figures

FourCastNeXt is an optimization of FourCastNet - a global machine learning weather forecasting model - that performs with a comparable level of accuracy and can be trained using around 5% of the original FourCastNet computational requirements. This technical report presents strategies for model optimization that maintain similar performance as measured by the root-mean-square error (RMSE) of the modelled variables. By providing a model with very low comparative training costs, FourCastNeXt makes Neural Earth System Modelling much more accessible to researchers looking to conduct training experiments and ablation studies. FourCastNeXt training and inference code are available at https://github.com/nci/FourCastNeXt, Comment: Major revision. All prior content (text, figures, table) has been updated. Additionally, new text, tables and figures have been added. Updated title. Updated author list

Image Quality Assessment (IQA) and Image Aesthetic Assessment (IAA) aim to simulate human subjective perception of image visual quality and aesthetic appeal. Existing methods typically address these tasks independently due to distinct learning objectives. However, they neglect the underlying interconnectedness of both tasks, which hinders the learning of task-agnostic shared representations for human subjective perception. To confront this challenge, we propose Unified vision-language pre-training of Quality and Aesthetics (UniQA), to learn general perceptions of two tasks, thereby benefiting them simultaneously. Addressing the absence of text in the IQA datasets and the presence of textual noise in the IAA datasets, (1) we utilize multimodal large language models (MLLMs) to generate high-quality text descriptions; (2) the generated text for IAA serves as metadata to purify noisy IAA data. To effectively adapt the pre-trained UniQA to downstream tasks, we further propose a lightweight adapter that utilizes versatile cues to fully exploit the extensive knowledge of the pre-trained model. Extensive experiments demonstrate that our approach attains a new state-of-the-art performance on both IQA and IAA tasks, while concurrently showcasing exceptional zero-shot and few-label image assessment capabilities. The source code will be available at https://github.com/zht8506/UniQA.

Remote Sensing Image-Text Retrieval (RSITR) is pivotal for knowledge services and data mining in the remote sensing (RS) domain. Considering the multi-scale representations in image content and text vocabulary can enable the models to learn richer representations and enhance retrieval. Current multi-scale RSITR approaches typically align multi-scale fused image features with text features, but overlook aligning image-text pairs at distinct scales separately. This oversight restricts their ability to learn joint representations suitable for effective retrieval. We introduce a novel Multi-Scale Alignment (MSA) method to overcome this limitation. Our method comprises three key innovations: (1) Multi-scale Cross-Modal Alignment Transformer (MSCMAT), which computes cross-attention between single-scale image features and localized text features, integrating global textual context to derive a matching score matrix within a mini-batch, (2) a multi-scale cross-modal semantic alignment loss that enforces semantic alignment across scales, and (3) a cross-scale multi-modal semantic consistency loss that uses the matching matrix from the largest scale to guide alignment at smaller scales. We evaluated our method across multiple datasets, demonstrating its efficacy with various visual backbones and establishing its superiority over existing state-of-the-art methods. The GitHub URL for our project is: https://github.com/yr666666/MSA, Comment: 16 pages, 9 figures

Traditional image-to-3D models often struggle with scenes containing multiple objects due to biases and occlusion complexities. To address this challenge, we present REPARO, a novel approach for compositional 3D asset generation from single images. REPARO employs a two-step process: first, it extracts individual objects from the scene and reconstructs their 3D meshes using off-the-shelf image-to-3D models; then, it optimizes the layout of these meshes through differentiable rendering techniques, ensuring coherent scene composition. By integrating optimal transport-based long-range appearance loss term and high-level semantic loss term in the differentiable rendering, REPARO can effectively recover the layout of 3D assets. The proposed method can significantly enhance object independence, detail accuracy, and overall scene coherence. Extensive evaluation of multi-object scenes demonstrates that our REPARO offers a comprehensive approach to address the complexities of multi-object 3D scene generation from single images.

Because of "the non-repeatability of the experiment settings and conditions" and "the variability of brain patterns among subjects", the data distributions across sessions and electrodes are different in cross-subject motor imagery (MI) studies, eventually reducing the performance of the classification model. Systematically summarised based on the existing studies, a novel temporal-electrode data distribution problem is investigated under both intra-subject and inter-subject scenarios in this paper. Based on the presented issue, a novel bridging domain adaptation network (BDAN) is proposed, aiming to minimise the data distribution difference across sessions in the aspect of the electrode, thus improving and enhancing model performance. In the proposed BDAN, deep features of all the EEG data are extracted via a specially designed spatial feature extractor. With the obtained spatio-temporal features, a special generative bridging domain is established, bridging the data from all the subjects across sessions. The difference across sessions and electrodes is then minimized using the customized bridging loss functions, and the known knowledge is automatically transferred through the constructed bridging domain. To show the effectiveness of the proposed BDAN, comparison experiments and ablation studies are conducted on a public EEG dataset. The overall comparison results demonstrate the superior performance of the proposed BDAN compared with the other advanced deep learning and domain adaptation methods.

Multi-modal behaviors exhibited by surrounding vehicles (SVs) can typically lead to traffic congestion and reduce the travel efficiency of autonomous vehicles (AVs) in dense traffic. This paper proposes a real-time parallel trajectory optimization method for the AV to achieve high travel efficiency in dynamic and congested environments. A spatiotemporal safety module is developed to facilitate the safe interaction between the AV and SVs in the presence of trajectory prediction errors resulting from the multi-modal behaviors of the SVs. By leveraging multiple shooting and constraint transcription, we transform the trajectory optimization problem into a nonlinear programming problem, which allows for the use of optimization solvers and parallel computing techniques to generate multiple feasible trajectories in parallel. Subsequently, these spatiotemporal trajectories are fed into a multi-objective evaluation module considering both safety and efficiency objectives, such that the optimal feasible trajectory corresponding to the optimal target lane can be selected. The proposed framework is validated through simulations in a dense and congested driving scenario with multiple uncertain SVs. The results demonstrate that our method enables the AV to safely navigate through a dense and congested traffic scenario while achieving high travel efficiency and task accuracy in real time.

Societal Impact Statement Despite comprising a small proportion of global agricultural land use, irrigated agriculture is enormously important to the global agricultural economy. Burgeoning food demand driven by population growth—together with reduced food supply caused by the climate crisis—is polarising the existing tension between water used for agricultural production versus that required for environmental conservation. We show that sustainable intensification via more diverse crop rotations, more efficient water application infrastructure and greater farm area under irrigation is conducive to greater farm business profitability under future climates. Summary Research aimed at improving crop productivity often does not account for the complexity of real farms underpinned by land-use changes in space and time. Here, we demonstrate how a new framework—WaterCan Profit—can be used to elicit such complexity using an irrigated case study farm with four whole-farm adaptation scenarios (Baseline, Diversified, Intensified and Simplified) with four types of irrigated infrastructure (Gravity, Pipe & Riser, Pivot and Drip). Without adaptation, the climate crisis detrimentally impacted on farm profitability due to the combination of increased evaporative demand and increased drought frequency. Whole-farm intensification—via greater irrigated land use, incorporation of rice, cotton and maize and increased nitrogen fertiliser application—was the only adaptation capable of raising farm productivity under future climates. Diversification through incorporation of grain legumes into crop rotations significantly improved profitability under historical climates; however, profitability of this adaptation declined under future climates. Simplified systems reduced economic risk but also had lower long-term economic returns. We conclude with four key insights: (1) When assessing whole-farm profit, metrics matter: Diversified systems generally had higher profitability than Intensifie

Si-Yu Chen,1 Rui-Lin Yang,1 Xiang-Chong Wu,2 De-Zhi Zhao,1 Sheng-Ping Fu,2 Feng-Qin Lin,1 Lin-Yan Li,1 Li-Mei Yu,1 Qian Zhang,3 Tao Zhang1,2 1Key Laboratory of Cell Engineering of Guizhou Province, Affiliated Hospital of Zunyi Medical University, Zunyi, Guizhou, People’s Republic of China; 2Department of Orthopaedic Surgery, Affiliated Hospital of Zunyi Medical University, Zunyi, Guizhou, People’s Republic of China; 3Department of Human Anatomy, Zunyi Medical University, Zunyi, Guizhou, People’s Republic of ChinaCorrespondence: Tao Zhang; Qian Zhang, Email oceanzt@163.com; zmczq@163.comAbstract: Spinal Cord Injury (SCI), with its morbidity characteristics of high disability rate and high mortality rate, is a disease that is highly destructive to both the physiology and psychology of the patient, and for which there is still a lack of effective treatment. Following spinal cord injury, a cascade of secondary injury reactions known as ischemia, peripheral inflammatory cell infiltration, oxidative stress, etc. create a microenvironment that is unfavorable to neural recovery and ultimately results in apoptosis and necrosis of neurons and glial cells. Mesenchymal stem cell (MSC) transplantation has emerged as a more promising therapeutic options in recent years. MSC can promote spinal cord injury repair through a variety of mechanisms, including immunomodulation, neuroprotection, and nerve regeneration, giving patients with spinal cord injury hope. In this paper, it is discussed the neuroprotection and nerve regeneration components of MSCs’ therapeutic method for treating spinal cord injuries.Keywords: spinal cord injury, mesenchymal stem cell, neuroinflammation, neuroprotection, nerve regeneration

Lan Gao,1 Lei-Lei Wang,1 Rui Yang,2 Xing-Jie Yang,1 Shuang-Jiang Zhou1 1Beijing HuiLongGuan Hospital, Peking University HuiLongGuan Clinical Medical School, Beijing, People’s Republic of China; 2Beijing Anding Hospital, Capital Medical University, Beijing, People’s Republic of ChinaCorrespondence: Shuang-Jiang Zhou, Beijing HuiLongGuan Hospital, Peking University HuiLongGuan Clinical Medical School, Beijing, 10096, People’s Republic of China, Tel +86-10-83024136, Email zhoushuangjiang@126.comPurpose: Depression and anxiety have a significant impact on an individuals’ work and personal life alike. The mental health of researchers is a significant concern worldwide. This study investigated the mental health status of Chinese researchers specifically and explored the moderating effects of perceived stress on the influence of low self-accomplishment on anxiety and depression.Methods: The online survey platform “Survey Star” was used to create a questionnaire to be distributed to researchers, with 949 questionnaires retrieved. The general information questionnaire, 10-item Perceived Stress Scale (PSS-10), Generalized Anxiety Disorder-7 (GAD-7), Patient Health Questionnaire-9 (PHQ-9), and Maslach Burnout Inventory General Survey (MBI-GS) were used for this investigation. Pearson’s correlation analysis was performed to investigate correlations among the relevant variables. Model 8 of PROCESS 3.3 program was used to analyze the moderating effects of perceived stress.Results: Among the 949 participants, 570 (60.1%) reported symptoms of depression and 431 (45.4%) had symptoms of anxiety, with about one in six reporting symptoms of self-harm or suicidal ideation. Perceived stress was found to moderate the effect of low self-accomplishment on depression and anxiety.Conclusion: Here we show that researchers exhibit a high rate of depression and anxiety symptoms. Perceived stress is also shown to play a moderating role on the influence of low self-accomplishment on

Black soldier fly have been shown to be one of the optimal alternatives to fishmeal, but there are few reports on the effects of zymolytic black soldier fly larvae (ZBSFL) on the growth and digestion of crustaceans. An 8-week feeding trial was conducted to evaluate the effects of different replacement levels of ZBFLS on growth performance, body composition, and digestive enzyme activity of the mud crab larvae. Four diets were formulated by replacing fishmeal with 0%, 5%, 10% and 15% ZBSFL in the basal diet. Crab larvae were randomly divided into four groups of three replicates each and fed twice daily. The results showed that the SR of crab larvae was higher than that of the no-substitution group when the substitution rate reached 5% (P < 0.05). There was no significant change in SR when the substitution rate was further increased. Weight growth rate and Specific growth rate were similar, both highest at 10% substitution ratio. The crude protein content of whole crab larvae gradually increased as the proportion of FM substituted by ZBSFL increased. The lipid content of whole crab larvae in the 5% substitution ratio group was significantly higher than that in all other groups (P < 0.05). Meanwhile. The activities of amylase, protease and lipase gradually increased. In this experiment, when the percentage of ZBSFL substitution for FM reached 10%, its growth performance was optimal, with higher SR, less negative effects and more balanced indicators in all aspects. When the substitution rate was further increased, it might increase the digestive burden of the crab and negatively affect its growth.

Offline goal-conditioned RL (GCRL) offers a way to train general-purpose agents from fully offline datasets. In addition to being conservative within the dataset, the generalization ability to achieve unseen goals is another fundamental challenge for offline GCRL. However, to the best of our knowledge, this problem has not been well studied yet. In this paper, we study out-of-distribution (OOD) generalization of offline GCRL both theoretically and empirically to identify factors that are important. In a number of experiments, we observe that weighted imitation learning enjoys better generalization than pessimism-based offline RL method. Based on this insight, we derive a theory for OOD generalization, which characterizes several important design choices. We then propose a new offline GCRL method, Generalizable Offline goAl-condiTioned RL (GOAT), by combining the findings from our theoretical and empirical studies. On a new benchmark containing 9 independent identically distributed (IID) tasks and 17 OOD tasks, GOAT outperforms current state-of-the-art methods by a large margin. © 2023 Proceedings of Machine Learning Research. All rights reserved.

MeV nuclear de-excitation lines serve as a unique tool to study low-energy cosmic rays (CRs), containing both spectral and elemental information of the interacting material. In this paper, we estimated the possible nuclear de-excitation lines from the young supernova remnant Cassiopeia A. Given different CR spectral shapes and interacting materials, we found the predicted fluxes of strong narrow line emissions from the remnant are highly model-dependent, ranging from about $1\times10^{-10}\,{\rm \,cm^{-2}\,s^{-1}}$ to $1\times10^{-6}\, {\rm \,cm^{-2}\,s^{-1}}$ for the 4.44 MeV narrow line and from about $4\times10^{-11}\,{\rm \,cm^{-2}\,s^{-1}}$ to $2\times10^{-7}{\rm \,cm^{-2}\,s^{-1}}$ for the 6.13 MeV narrow line, respectively. Based on the new estimation, we also discussed the detection probability of these line emissions against the MeV diffuse Galactic background under different assumptions of instrument response functions., Comment: 6 pages, 6 figures, accepted for publication in MNRAS

Goal-conditioned reinforcement learning (RL) with sparse rewards remains a challenging problem in deep RL. Hindsight Experience Replay (HER) has been demonstrated to be an effective solution, where HER replaces desired goals in failed experiences with practically achieved states. Existing approaches mainly focus on either exploration or exploitation to improve the performance of HER. From a joint perspective, exploiting specific past experiences can also implicitly drive exploration. Therefore, we concentrate on prioritizing both original and relabeled samples for efficient goal-conditioned RL. To achieve this, we propose a novel value consistency prioritization (VCP) method, where the priority of samples is determined by the consistency of ensemble Q-values. This distinguishes the VCP method with most existing prioritization approaches which prioritizes samples based on the uncertainty of ensemble Q-values. Through extensive experiments, we demonstrate that VCP achieves significantly higher sample efficiency than existing algorithms on a range of challenging goal-conditioned manipulation tasks. We also visualize how VCP prioritizes good experiences to enhance policy learning.

Tong Zhang,&ast; Rui Yang,&ast; Juhua Pan, Shijing Huang Research and Development Center of Traditional Chinese Medicine, Guang’anmen Hospital, China Academy of Chinese Medical Sciences, Beijing, People’s Republic of China&ast;These authors contributed equally to this workCorrespondence: Shijing Huang; Juhua Pan, Guang’anmen Hospital, China Academy of Chinese Medical Sciences, Beijing, 100053, People’s Republic of China, Email gamhsj@126.com; panjuhua@126.comBackground: Parkinson’s disease (PD) is one of the common neurodegenerative diseases. Depression and anxiety are the most common psychiatric symptoms of PD. It is important to study the potential relationship between PD and depression or anxiety.Aim: This study aimed to use bibliometrics to analyze the papers about parkinson’s disease related depression and anxiety over the last 22 years, and to characterize the current status of research and predict future hotspots.Methods: In the Web of Science Core Collection (WoSCC) from 2000 to 2022, documents are searched according to specific subject words. The selected literature was retrospectively analyzed and mapped using CiteSpace and Vosviewer software. We analyzed countries, institutions, journals, authors, references and keywords.Results: A total of 7368 papers were included from 2000 to 2022, and the number of publications has shown an upward trend year by year. Movement Disorder is the journal with the highest number of publications (391 publications, 5.31%) and citations (30,549 times), with the United States (2055 publications, 27.9%) and the University of Toronto (158 publications) being the countries and institutions with the highest number of publications. The high-frequency keywords focused on “quality of life”, “deep brain stimulation” and “non-motor symptoms”. “Functional connectivity”, “gut microbiota” and “inflammation” may be at the forefront of future research.Conclusion: Parkinson’s disease related d

Multi-modal representation methods have achieved advanced performance in medical applications by extracting more robust features from multi-domain data. However, existing methods usually need to train additional branches for downstream tasks, which may increase the model complexities in clinical applications as well as introduce additional human inductive bias. Besides, very few studies exploit the rich clinical knowledge embedded in clinical daily reports. To this end, we propose a novel medical generalist agent, MGA, that can address three kinds of common clinical tasks via clinical reports knowledge transformation. Unlike the existing methods, MGA can easily adapt to different tasks without specific downstream branches when their corresponding annotations are missing. More importantly, we are the first attempt to use medical professional language guidance as a transmission medium to guide the agent's behavior. The proposed method is implemented on four well-known X-ray open-source datasets, MIMIC-CXR, CheXpert, MIMIC-CXR-JPG, and MIMIC-CXR-MS. Promising results are obtained, which validate the effectiveness of our proposed MGA. Code is available at: https://github.com/SZUHvern/MGA

Interactive devices with touch screen have become commonly used in various aspects of daily life, which raises the demand for high production quality of touch screen glass. While it is desirable to develop effective defect detection technologies to optimize the automatic touch screen production lines, the development of these technologies suffers from the lack of publicly available datasets. To address this issue, we in this paper propose a dedicated touch screen glass defect dataset which includes seven types of defects and consists of 2504 images captured in various scenarios.All data are captured with professional acquisition equipment on the fixed workstation. Additionally, we benchmark the CNN- and Transformer-based object detection frameworks on the proposed dataset to demonstrate the challenges of defect detection on high-resolution images. Dataset and related code will be available at https://github.com/Yangr116/SSGDataset.

While mesh networking for edge settings (e.g., smart buildings, farms, battlefields, etc.) has received much attention, the layer of control over such meshes remains largely centralized and cloud-based. This paper focuses on applications with sense-trigger-actuate (STA) workloads -- these are similar to the abstraction of routines popular in smart homes, but applied to larger-scale edge IoT deployments. We present CoMesh, which tackles the challenge of building local, non-cloud, and decentralized solutions for control of sense-trigger-actuate applications. At its core CoMesh uses an abstraction called k-groups to spread in a fine-grained way, the load of STA actions. Coordination within the k-group uses selective fast and cheap mechanisms rather than expensive off-the-shelf solutions. k-group selection is proactively dynamic, and occurs by using a combination of zero-message-exchange mechanisms (to reduce load) and locality sensitive hashing (to be aware of physical layout of devices). We analyze and theoretically prove the safety of CoMesh's mechanisms. Our evaluations using both simulation and Raspberry Pi lab deployments show that CoMesh is load-balanced, fast, and fault-tolerant., Comment: 12 pages, 12 figures

Recent work has shown that representation learning plays a critical role in sample-efficient reinforcement learning (RL) from pixels. Unfortunately, in real-world scenarios, representation learning is usually fragile to task-irrelevant distractions such as variations in background or viewpoint. To tackle this problem, we propose a novel clustering-based approach, namely Clustering with Bisimulation Metrics (CBM), which learns robust representations by grouping visual observations in the latent space. Specifically, CBM alternates between two steps: (1) grouping observations by measuring their bisimulation distances to the learned prototypes; (2) learning a set of prototypes according to the current cluster assignments. Computing cluster assignments with bisimulation metrics enables CBM to capture task-relevant information, as bisimulation metrics quantify the behavioral similarity between observations. Moreover, CBM encourages the consistency of representations within each group, which facilitates filtering out task-irrelevant information and thus induces robust representations against distractions. An appealing feature is that CBM can achieve sample-efficient representation learning even if multiple distractions exist simultaneously.Experiments demonstrate that CBM significantly improves the sample efficiency of popular visual RL algorithms and achieves state-of-the-art performance on both multiple and single distraction settings. The code is available at https://github.com/MIRALab-USTC/RL-CBM., Comment: Accepted to AAAI 2023

Generalization in partially observed markov decision processes (POMDPs) is critical for successful applications of visual reinforcement learning (VRL) in real scenarios. A widely used idea is to learn task-relevant representations that encode task-relevant information of common features in POMDPs, i.e., rewards and transition dynamics. As transition dynamics in the latent state space -- which are task-relevant and invariant to visual distractions -- are unknown to the agents, existing methods alternatively use transition dynamics in the observation space to extract task-relevant information in transition dynamics. However, such transition dynamics in the observation space involve task-irrelevant visual distractions, degrading the generalization performance of VRL methods. To tackle this problem, we propose the reward sequence distribution conditioned on the starting observation and the predefined subsequent action sequence (RSD-OA). The appealing features of RSD-OA include that: (1) RSD-OA is invariant to visual distractions, as it is conditioned on the predefined subsequent action sequence without task-irrelevant information from transition dynamics, and (2) the reward sequence captures long-term task-relevant information in both rewards and transition dynamics. Experiments demonstrate that our representation learning approach based on RSD-OA significantly improves the generalization performance on unseen environments, outperforming several state-of-the-arts on DeepMind Control tasks with visual distractions., Comment: 23 pages

Lan Gao,1,&ast; Rui Yang,2,&ast; Hong-Zhen Fan,1 Lei-Lei Wang,1 Yan-Li Zhao,1 Shu-Ping Tan,1 Chun-Ling Xiao,1 Shuang-Jiang Zhou1 1Beijing HuiLongGuan Hospital, Peking University HuiLongGuan Clinical Medical School, Beijing, People’s Republic of China; 2Beijing Anding Hospital, Capital Medical University, Beijing, People’s Republic of China&ast;These authors contributed equally to this workCorrespondence: Chun-Ling Xiao; Shuang-Jiang Zhou, Beijing HuiLongGuan Hospital, Peking University HuiLongGuan Clinical Medical School, Beijing, 10096, People’s Republic of China, Email xiaocl2002@163.com; zhoushuangjiang@126.comPurpose: To explore the correlation between aggressive behavior and impulsive and aggressive personality traits in inpatients with schizophrenia.Methods: In total, 367 inpatients with schizophrenia were divided into two groups: the aggressive group and the non-aggressive group. We assessed inpatients’ psychotic symptoms as well as their aggressive and impulsive personality traits using the Positive and Negative Symptom Scale, the Barratt Impulsiveness Scale, and the Buss-Perry Aggression Questionnaire.Results: Compared with the scores of inpatients in the non-aggressive group, the total Buss-Perry Aggression Questionnaire, subscale, and Barratt Impulsiveness Scale behavioral factor scores in those in the aggressive group were higher (p < 0.05). The results of logistic regression analysis suggested that a high Positive and Negative Symptom Scale positive factor score (odds ratio = 1.07) and a high Buss-Perry Aggression Questionnaire physical aggression score (odds ratio = 1.02) were risk factors for aggressive behavior.Conclusion: Hospitalized patients with schizophrenia with more severe positive symptoms and aggressive traits may be more prone to aggressive behavior.Keywords: schizophrenia, aggressive behavior, impulsivity trait, aggressive trait, stable patients

Shuang-Jiang Zhou,1,&ast; Rui Yang,2,&ast; Lei-Lei Wang,1 Meng Qi,1 Xiao-Fei Yuan,2 Ting-Ting Wang,3 Tian-He Song,4 Yun-Yue Zhuang,4 Hong-Juan Li,1 Yun-Long Tan,1 Xue Wang,2 Jing-Xu Chen1 1Sleep Medicine Center, Beijing HuiLongGuan Hospital, Peking University HuiLongGuan Clinical Medical School, Beijing, People’s Republic of China; 2Beijing Anding Hospital, Capital Medical University, Beijing, People’s Republic of China; 3School of Mental Health, Bengbu Medical College, Bengbu, People’s Republic of China; 4Department of Psychology, Chengde Medical University, Chengde, People’s Republic of China&ast;These authors contributed equally to this workCorrespondence: Xue Wang, Beijing Anding Hospital, Capital Medical University, Beijing, 100088, People’s Republic of China, Tel +86-10-58303034, Email xuew37@ccmu.edu.cn Jing-Xu Chen, Beijing HuiLongGuan Hospital, Peking University HuiLongGuan Clinical Medical School, Beijing, 100096, People’s Republic of China, Tel +86-10-83024278, Email chenjx1110@163.comObjective: To assess the performance of a wearable multi-sensor system (SensEcho) in comparison to polysomnography (PSG) in measuring sleep stages and searching for obstructive sleep apnea (OSA).Methods: Participants underwent overnight simultaneous monitoring using SensEcho and PSG in a sleep laboratory. SensEcho analyzed the recordings spontaneously, and PSG was assessed as per standard guidelines. The degree of snoring was evaluated according to the guidelines for the diagnosis and treatment of OSA hypopnea syndrome (2011 revision). The Epworth Sleepiness Scale (ESS) was used to assess general daytime sleepiness.Results: This study included 103 Han Chinese, 91 of whom (age 39.02 ± 13.84 years, body mass index 27.28 ± 5.12 kg/m2, 61.54% male) completed the assessments. The measures of total sleep time (P = 0.198); total wake time (P = 0.182); shallow sleep (P = 0.297), deep sleep (P = 0.422), rapid eye movement sleep (P = 0.570), and awake (P = 0.

This paper aims to efficiently enable Large Language Models (LLMs) to use multimodal tools. Advanced proprietary LLMs, such as ChatGPT and GPT-4, have shown great potential for tool usage through sophisticated prompt engineering. Nevertheless, these models typically rely on prohibitive computational costs and publicly inaccessible data. To address these challenges, we propose the GPT4Tools based on self-instruct to enable open-source LLMs, such as LLaMA and OPT, to use tools. It generates an instruction-following dataset by prompting an advanced teacher with various multi-modal contexts. By using the Low-Rank Adaptation (LoRA) optimization, our approach facilitates the open-source LLMs to solve a range of visual problems, including visual comprehension and image generation. Moreover, we provide a benchmark to evaluate the ability of LLMs to use tools, which is performed in both zero-shot and fine-tuning ways. Extensive experiments demonstrate the effectiveness of our method on various language models, which not only significantly improves the accuracy of invoking seen tools, but also enables the zero-shot capacity for unseen tools. The code and demo are available at https://github.com/StevenGrove/GPT4Tools.

A major problem with Active Learning (AL) is high training costs since models are typically retrained from scratch after every query round. We start by demonstrating that standard AL on neural networks with warm starting fails, both to accelerate training and to avoid catastrophic forgetting when using fine-tuning over AL query rounds. We then develop a new class of techniques, circumventing this problem, by biasing further training towards previously labeled sets. We accomplish this by employing existing, and developing novel, replay-based Continual Learning (CL) algorithms that are effective at quickly learning the new without forgetting the old, especially when data comes from an evolving distribution. We call this paradigm Continual Active Learning (CAL). We show CAL achieves significant speedups using a plethora of replay schemes that use model distillation and that select diverse, uncertain points from the history. We conduct experiments across many data domains, including natural language, vision, medical imaging, and computational biology, each with different neural architectures and dataset sizes. CAL consistently provides a 3x reduction in training time, while retaining performance., Comment: Appeared in TMLR 2023

Due to the limitations of inadequate Whole-Slide Image (WSI) samples with weak labels, pseudo-bag-based multiple instance learning (MIL) appears as a vibrant prospect in WSI classification. However, the pseudo-bag dividing scheme, often crucial for classification performance, is still an open topic worth exploring. Therefore, this paper proposes a novel scheme, ProtoDiv, using a bag prototype to guide the division of WSI pseudo-bags. Rather than designing complex network architecture, this scheme takes a plugin-and-play approach to safely augment WSI data for effective training while preserving sample consistency. Furthermore, we specially devise an attention-based prototype that could be optimized dynamically in training to adapt to a classification task. We apply our ProtoDiv scheme on seven baseline models, and then carry out a group of comparison experiments on two public WSI datasets. Experiments confirm our ProtoDiv could usually bring obvious performance improvements to WSI classification., Comment: 12 pages, 5 figures, and 3 tables

Box-supervised instance segmentation has gained much attention as it requires only simple box annotations instead of costly mask or polygon annotations. However, existing box-supervised instance segmentation models mainly focus on mask-based frameworks. We propose a new end-to-end training technique, termed BoxSnake, to achieve effective polygonal instance segmentation using only box annotations for the first time. Our method consists of two loss functions: (1) a point-based unary loss that constrains the bounding box of predicted polygons to achieve coarse-grained segmentation; and (2) a distance-aware pairwise loss that encourages the predicted polygons to fit the object boundaries. Compared with the mask-based weakly-supervised methods, BoxSnake further reduces the performance gap between the predicted segmentation and the bounding box, and shows significant superiority on the Cityscapes dataset. The code has been available publicly., Comment: ICCV 2023

The density of cosmic rays inside molecular clouds determines the ionization rate in the dense cores where stars form. It is also one of the drivers of astrochemistry leading to the creation of complex molecules. Through Fermi Large Area Telescope observations of nearby giant molecular clouds, we observed deficits (holes) in the gamma-ray residual map when modelling with the expected gamma-ray diffuse emission from uniform cosmic rays interacting with the molecular content. We propose that the deficit is due to the lack of penetration of the low-energy (sub-GeV to GeV) cosmic rays into denser regions or clumps. This differs from the prevailing view of fast cosmic ray transport in giant molecular clouds where the magnetic turbulence is suppressed by neutral-ion damping, as our results require a slow diffusion inside dense molecular clumps. Through modelling we find that while the shielding is negligible on the cloud scale, it becomes important in the denser, parsec-sized regions where the gravitational collapse is already at play, changing the initial condition of star formation and astrochemistry., Comment: Published in Nature Astronomy