Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks349 from public data
- Automatically Designing CNN Architectures Using the Genetic Algorithm for Image Classification879
This article proposes an automatic CNN architecture design method by using genetic algorithms, to effectively address the image classification tasks and shows the very comparable classification accuracy to the best one from manually designed and automatic + manually tuning CNNs, while consuming fewer computational resources.
- TAT-QA: A Question Answering Benchmark on a Hybrid of Tabular and Textual Content in Finance652
This work extracts samples from real financial reports to build a new large-scale QA dataset containing both Tabular And Textual data, named TAT-QA, where numerical reasoning is usually required to infer the answer, such as addition, subtraction, multiplication, division, counting, comparison/sorting, and the compositions.
- 650
- 472
- COMIC: Multi-view Clustering Without Parameter Selection369
A novel objective function to project raw data into one space in which the projection embraces the geometric consistency (GC) and the cluster assignment consistency (CAC) and could achieve promising performance in learning view-specific representation and eliminating the heterogeneous gaps across different views.
- 366
- 279
- Communication-Efficient Federated Learning With Compensated Overlap-FedAvg220
Overlap-FedAvg is an innovative framework that loosed the chain-like constraint of federated learning and paralleled the model training phase with the model communication phase (i.e., uploading local models and downloading the global model), so that the latter phase could be totally covered by the former phase.
- 208
- Learning Event-Based Motion Deblurring200
This paper starts from a sequential formulation of event-based motion deblurring, then shows how its optimization can be unfolded with a novel end-toend deep architecture, and proposes a differentiable directional event filtering module to effectively extract rich boundary prior from the evolution of events.
- Multi-view Spectral Clustering Network156
To deeply cluster multi-view data, MvSCN incorporates the local invariance within every single view and the consistency across different views into a novel objective function, where theLocal invariance is defined by a deep metric learning network rather than the Euclidean distance adopted by traditional approaches.
- Partially View-aligned Clustering151
A novel multi-view clustering method termed partially view-aligned clustering (PVC), which proposes to use a differentiable surrogate of the non-differentiable Hungarian algorithm and recasts it as a pluggable module to establish the category-level correspondence of the unaligned data.
- 110
- 97
- 93
- 81
- A Deep Learning-Based End-to-End Composite System for Hand Detection and Gesture Recognition79
This work proposes a deep learning-based architecture to jointly detect and classify hand gestures and demonstrates that the proposed architecture is efficient and robust, and outperforms other approaches in both the hand detection and gesture classification tasks.
- 76
- GLGE: A New General Language Generation Evaluation Benchmark75
The General Language Generation Evaluation (GLGE), a new multi-task benchmark for evaluating the generalization capabilities of NLG models across eight language generation tasks, is presented and a leaderboard with strong baselines including MASS, BART, and ProphetNet is built.
- 71
- Unsupervised Multi-Manifold Clustering by Learning Deep Representation.71
A novel deep manifold clustering method for learning effective deep representations and partitioning a dataset into clusters where each cluster contains data points from a single nonlinear manifold that can be intuitively extended to cluster out-of-sample datum.
- 68
- Partial Differential Equations Meet Deep Neural Networks: A Survey60
This survey fills the gap in a systematic review of the diverse neural network approaches for PDEs by categorizing and reviewing the current progress of deep NNs (DNNs) for PDEs, and offers a broader taxonomy and analyze applications across scientific, engineering, and medical fields.
- 59
- RikiNet: Reading Wikipedia Pages for Natural Question Answering59
This paper introduces a new model, called RikiNet, which reads Wikipedia pages for natural question answering, which is the first single model that outperforms the single human performance.
- Learning Inverse Mapping by AutoEncoder Based Generative Adversarial Nets59
A new approach based on using inverse generator (IG) model as encoder and pre-trained generator (G) as decoder of an AutoEncoder network to train the IG model, which can overcome the difficulty in training and inverse model of an non one-to-one function.
- 57
- Adaptive local structure learning for document co-clustering57
This paper proposes a new framework of co-clustering with adaptive local structure learning, based on nonnegative matrix tri-factorization, that performs intrinsic structure learning and 3-factor factorization simultaneously and explores not only the adaptiveLocal structure of the data space, but also the adaptive localructure of the feature space.
- 55
- 54
- 54
- 53
- Model-based evaluation of alternative reactive class closure strategies against COVID-1953
It is shown that, compared to symptom-prompted PCR testing, screening for cases in schools with antigen tests leads to greater reductions in infection rates in both students and the wider community.
- Multi-Scale Adaptive Network for Single Image Denoising53
This paper reveals the missing piece for multi-scale architecture design and proposes a novel Multi-Scale Adaptive Network (MSANet) for single image denoising, which simultaneously embraces the within-scale characteristics and the cross-scale complementarity.
- Revision in Continuous Space: Unsupervised Text Style Transfer without Adversarial Learning53
A new framework that utilizes the gradients to revise the sentence in a continuous space during inference to achieve text style transfer and significantly outperforms five state-of-the-art methods.
- 51
- 51
- 50
- Discriminative globality and locality preserving graph embedding for dimensionality reduction50
This article proposes a novel graph-based dimensionality reduction method entitled discriminative globality and locality preserving graph embedding (DGLPGE) by designing the informative globalities and locality preserves graph constructions.
- 49
- Multi-model ensemble gesture recognition network for high-accuracy dynamic hand gesture recognition49
A deep ensemble framework called multi-model ensemble gesture recognition network (MMEGRN) for skeleton-based hand gesture recognition, consisting of four sub-networks, three spatio-temporal features classifiers to leverage their various capabilities of extracting and classifying skeleton sequences.
- μ-Forcing49
The proposed method directly injects extra constraints on the posteriors of latent variables into the learning process of VRAE, which can flexibly and stably control the tradeoff between the Kullback-Leibler (KL) term and the reconstruction term, making the model learn dense and meaningful latent representations.
- Image Clustering with External Guidance48
An externally guided clustering method (Text-Aided Clustering, TAC), which leverages the textual semantics of WordNet to facilitate image clustering, which achieves state-of-the-art performance on five widely used and three more challengingimage clustering benchmarks, including the full ImageNet-1K dataset.
- 48
- The Correlation of Hippocampal T2-Mapping with Neuropsychology Test in Patients with Alzheimer’s Disease47
A positive correlation between T2-values and cognition scores suggests that quantitative T2 can be used in the early diagnosis of AD and in the monitoring of the treatment response.
- 46
- AnchiBERT: A Pre-Trained Model for Ancient Chinese Language Understanding and Generation42
AnchiBERT is released, a pre-trained language model based on the architecture of BERT, which is trained on large-scale ancient Chinese corpora and outperforms BERT as well as the non-pretrained models and achieves state-of-the-art results in all cases.
- 40
- 40
- End-to-end heart sound segmentation using deep convolutional recurrent network40
A novel end-to-end method based on convolutional long short-term memory (CLSTM), which directly uses audio recording as input to address HSS tasks and can be regarded as a powerful feature extraction tool, which can be integrated into the existing models for HSS.
- Tell Me How to Ask Again: Question Data Augmentation with Controllable Rewriting in Continuous Space40
A novel data augmentation method, referred to as Controllable Rewriting based Question Data Augmentation (CRQDA), for machine reading comprehension, question generation, and question-answering natural language inference tasks.
- Metric Multi-View Graph Clustering38
This work proposes to seamlessly integrates metric learning and graph learning for multi-view clustering, and employs a useful metric to depict the inherent structure with linearity-aware of affinity graph representation learned based on the self-expressiveness property.
- 35
- Heart sound segmentation via Duration Long–Short Term Memory neural network34
A Duration Long–Short Term Memory network (Duration LSTM) is proposed to effectively address the problem of poor utilization of the duration information of heart cycles by incorporating the duration features.
- Two-phase probabilistic collaborative representation-based classification34
The coarse to fine representation is adopted to propose two-phase probabilistic collaborative representation based-classification (TPCRC) to enhance the power of pattern discrimination in PCRC and the proposed methods outperform the state-of-the-art collaborative representation-based classification methods.
- 33
- Cataract detection based on ocular B-ultrasound images by collaborative monitoring deep learning32
An object detection network based on YOLO-v3 is introduced and classification loss with the aid of collaborative loss, which distinguishes whether the eyeball and lens belong to the same category, improves the classification accuracy in cataract detection.
- A Multi-Modal Chinese Poetry Generation Model31
A three-stage multi-modal Chinese poetry generation approach which can effectively capture character, phrase, and sentence information between contexts and improve the symmetry delivered in poems is proposed.
- Federated CINN Clustering for Accurate Clustered Federated Learning30
The Federated cINN Clustering Algorithm (FCCA) is proposed to robustly cluster clients into different groups, avoiding mutual interference between clients with data heterogeneity, and thereby enhancing the performance of the global model.
- 30
- Interacting with Non-Cooperative User30
A new solution named I-Pro is proposed that can learn Proactive policy in the Interactive setting via a learned goal weight, which consists of four factors (dialogue turn, goal completion difficulty, user satisfaction estimation, and cooperative degree).
- 29
- Multi-View Clustering on Topological Manifold29
This paper coalesces multiple view-wise graphs with the topological relevance considered, and learns the weights as well as the consensus graph interactively in a unified framework, and manipulates the consensusgraph by a connectivity constraint to exploit the implied data manifold.
- Learning Smooth Representation for Multi-view Subspace Clustering29
This paper proposes to achieve a smooth representation for each view and thus facilitate the downstream clustering task and achieves the smooth representation learning as well as multi-view clustering interactively in a unified framework, hence it is an end-to-end single-stage learning problem.
- Deep Fair Clustering via Maximizing and Minimizing Mutual Information: Theory, Algorithm and Metric28
This work develops a mutual information theory for deep fair clustering and accordingly designs a novel algorithm, dubbed FCMI, designed to achieve four characteristics highly expected by deepFair clustering, i.e., compact, balanced, and fair clusters, as well as informative features.
- Zero-shot learning by mutual information estimation and maximization28
This paper proposes to build the visual-semantic embedding by maximizing the mutual information between visual features and corresponding attributes and introduces the noise-contrastive estimation to calculate lower-bound value of mutual information.
- 28
- Anesthetic drug discovery with computer-aided drug design and machine learning27
A review of the computational methods employed in the development of novel anesthetics, outlining their respective advantages and limitations aims to help relevant researchers develop safer and more effective anesthetic drugs.
- 27
- Image segmentation of nasopharyngeal carcinoma using 3D CNN with long-range skip connection and multi-scale feature pyramid27
A 3D convolutional neural network with long-range skip connection and multi-scale feature pyramid and multi-scale feature pyramid for the segmentation of images of NPC, which is superior to five state-of-the-art networks and equivalent to the judgment of an experienced physician.
- Crowdfunding mechanism comparison if there are altruistic donors27
Study of how the crowdfunding price, funding goal, and mechanism selection are influenced by the participation of altruistic donors who contribute money to help a crowdfunding campaign reach its goal shows that when the creator raises funds only through crowdfunding, the contributions from donors encourage the creator to choose AON, while when thecreator will approach a venture capitalist for further investment after crowdfunding, donor contributions encourage thecreator to choose KIA.
- 26
- DrugLLM: Open Large Language Model for Few-shot Molecule Generation25
This work introduced DrugLLM, a LLM tailored for drug design that learns how to modify molecules in drug discovery by predicting the next molecule based on past modifications, presenting a powerful few-shot molecule generation capacity.
- A New Discriminative Collaborative Neighbor Representation Method for Robust Face Recognition25
A new discriminative collaborative neighbor representation (DCNR) method for face recognition is proposed by integrating class discrimination and data locality, and it is shown that the proposed methods achieve competitive classification performance, compared to the state-of-the-art representation-based classification methods.
- Diffusion Posterior Proximal Sampling for Image Restoration24
This paper opts for a sample consistent with the measurement identity at each generative step, exploiting the sampling selection as an avenue for output stability and enhancement, and presents a refined paradigm for diffusion-based image restoration.
- The COVID-19 outbreak in Sichuan, China: Epidemiology and impact of interventions24
The findings indicate that, were the control measures be adopted four weeks later, the epidemic could have lasted 49 days longer (95%CI: 31-68 days), causing 9,216 more cases and possibly overwhelming Sichuan healthcare system.
- 24
- 23
- 23
- CDD: Multi-view Subspace Clustering via Cross-view Diversity Detection23
A novel multi-view subspace clustering via cross-view diversity detection (CDD) is lodge that outperforms several other state-of-the-art methods and is able to exploit these two complementary criteria seamlessly into a holistic design of clustering algorithms.
- 22
- 22
- Face inpainting based on GAN by facial prediction and fusion as guidance information22
A new generative adversarial network by facial prediction and fusion as guidance information, is proposed for large missing regions of face inpainting to fully explore face region semantic information and generate coordinated coarse face images.
- 22
- Diverse, Controllable, and Keyphrase-Aware: A Corpus and Method for News Multi-Headline Generation22
A multi-source Transformer decoder, which takes three sources as inputs and generates multiple headlines with keyphrases of user interests to generate multiple keyphrase-relevant headlines, achieves state-of-the-art results in terms of quality and diversity.
- 21
- 21
- 21
- An Abstract Painting Generation Method Based on Deep Generative Model21
A novel method of generating abstract paintings is proposed using the public painting dataset WikiArt and a K-Means algorithm that automatically finds the optimal K value to perform color segmentation on these images, and divides the picture into different color blocks.
- 20
- Ancient–Modern Chinese Translation with a New Large Training Dataset20
This article proposes an ancient–modern Chinese clause alignment approach based on the characteristics of these two languages that achieves 94.2 F1-score on the authors' manual annotation Test set and creates a new large-scale ancient– modern Chinese parallel corpus that contains 1.24M bilingual pairs.
- Concept -- An Evaluation Protocol on Conversational Recommender Systems with System-centric and User-centric Factors19
A new and inclusive evaluation protocol, Concept, which integrates both system- and user-centric factors in CRS, and pinpoints the problem of low usability in the ChatGPT and offers a comprehensive reference guide for evaluating CRS, thereby setting the foundation for CRS improvement.
- GPT-NAS: Evolutionary Neural Architecture Search with the Generative Pre-Trained Model19
This work presents a novel architecture search al-gorithm, called GPT-NAS, that optimizes neural architectures by Generative Pre-Trained (GPT) model with an evolutionary algorithm (EA) as the search strategy.
- 19
- Shortest path computation using pulse-coupled neural networks with restricted autowave19
This paper develops a Pulse-Coupled Neural Network model to efficiently compute a single-pair shortest path, which is endowed with a special mechanism, called on-forward/off-backward, which significantly reduces the computational cost of searching for the shortest path.
- Triple-Scale Structure-Induced Efficient Passive Radiative Cooling Combining Robust Anticondensation18
- 18
- Deep Density-Aware Count Regressor18
This work introduces multilayer gradient fusion for training a density-aware global count regressor and improves benchmark results on public datasets and exhibits itself to be a new solution to crowd counting problems in practice.
- An Automatic Registration Algorithm for 3D Maxillofacial Model18
The proposed registration algorithm can achieve a good alignment result between partial and whole maxillofacial model in spite of ambiguous matching, which has a potential application in the oral and maxill ofacial reparative and reconstructive surgery.
- Multi‐style Chinese art painting generation of flowers17
Extensive experiments show that the proposed Flower-Generative Adversarial Network framework can produce better and multi-style Chinese art painting of flowers than existing methods.
- Dual-modality visual feature flow for medical report generation16
Dual-Modality Visual Feature Flow (DMVF) is proposed, which introduces region-level features based on grid-level features to enhance the method's ability to identify lesions and key areas and enhances two types of feature flows based on their attributes to prevent the loss of key information.
- Dual-feature Fusion Attention Network for Small Object Segmentation16
A dual feature fusion attention network (DFF-Net) to improve the segmentation accuracy of small objects and has fewer parameters, faster inference, and lower model complexity, and achieves better accuracy than more state-of-the-art methods.
- POS-Constrained Parallel Decoding for Non-autoregressive Generation16
The proposed POS-constrained Parallel Decoding (POSPD) method aims at providing a specific POS sequence to constrain the NAG model during decoding, and demonstrates that POSPD consistently improves NAG models on four text generation tasks to a greater extent compared to knowledge distillation.
- 15
- 15
- Lifelong Multi-view Spectral Clustering15
This paper uses view-specific knowledge libraries to capture intra-view knowledge across different tasks and proposes two types of libraries: an orthogonal basis library that stores cluster centers in consecutive tasks, and a feature embedding library that embeds feature relations shared among correlated tasks.
- Compression of deep neural networks: bridging the gap between conventional-based pruning and evolutionary approach15
A novel structured pruning method called Conventional-based and Evolutionary Approaches Guided Multiobjective Pruning (CEA-MOP) is proposed to address this problem, where the power of conventional pruning methods is effectively exploited for the evolutionary process.
- A fully automated sex estimation for proximal femur X-ray images through deep learning detection and classification15
The model based on convolutional neural networks has an accuracy similar to that of current state-of-the-art mathematical functions using manually extracted features for the Chinese Han population samples, proving to be a reliable choice for the human sex estimation.
- 15
- 14
- DD-Ranking: Rethinking the Evaluation of Dataset Distillation14
DD-Ranking, a unified evaluation framework, along with new general evaluation metrics to uncover the true performance improvements achieved by different methods are proposed, which provide a more comprehensive and fair evaluation standard for future research advancements.
- Communication-efficient Federated Learning with Single-Step Synthetic Features Compressor for Faster Convergence14
Experiments on multiple datasets and models suggest that 3SFC has significantly better convergence rates compared to competing methods with lower compression rates and ablation studies and visualizations show that 3SFC can carry more information than competing methods for every communication round, further validating its effectiveness.
- 14
- 14
- Heart-Darts: Classification of Heartbeats Using Differentiable Architecture Search14
A novel approach to efficiently classify the ECG signals by automatically designing the CNN model with the differentiable architecture search (i.e., Darts, a cell-based neural architecture search method) that outperforms several state-of-the-art CNN models in ECG classification in terms of both performance and generalization capability.
- Deep Poetry: A Chinese Classical Poetry Generation System14
This work demonstrates a Chinese classical poetry generation system called Deep Poetry that uses neural networks that are trained on over 200 thousand poems and 3 million ancient Chinese prose to generate classical poetry.
- Aesthetic assessment of paintings based on visual balance14
A novel method is proposed to quantify the law of visual balance and automatically evaluate the aesthetic value of images and it is demonstrated qualitatively and quantitatively that the proposed evaluation method is basically consistent with the intuitive experience of most human beings.
- An Improved Backpropagation Algorithm Using Absolute Error Function14
An improved backpropagation algorithm is proposed by using the Lyapunov method to minimize the absolution error function, which is more robust and faster for learning than the traditional square error function when target signals include some incorrect data.
- 13
- Unsupervised Neural Rendering for Image Hazing13
Compared with existing haze generation methods, HazeGEN renders the hazy images in an unsupervised, learnable, and controllable manner, thus avoiding the labor-intensive efforts in paired data collection and the domain-shift issue in haze generation.
- 13
- 13
- 13
- An angle-based method for measuring the semantic similarity between visual and textual features13
This paper learns the feature representation of images and their sentence descriptions by different deep neural networks to learn about the inner-modal correspondences between visual and language data, and develops the learning algorithm on neural networks for expressing the semantic features of texts and images.
- A neural networks learning algorithm for minor component analysis and its convergence analysis13
A neural networks learning algorithm for estimating adaptively minor component from input signals via a deterministic discrete time (DDT) method is proposed.
- 13
- Cross-Space Adaptive Filter: Integrating Graph Topology and Node Attributes for Alleviating the Over-smoothing Problem12
This paper proposes a cross-space adaptive filter, called CSF, to produce the adaptive-frequency information extracted from both the topology and attribute spaces, and derives the cross-space filter via an effective multiple-kernel learning strategy, which helps to address the over-smoothing problem while maintaining effectiveness.
- Revision in Continuous Space: Fine-Grained Control of Text Style Transfer.12
A new framework without adversarial learning is proposed, which consists of a variational auto-encoder (VAE), some attribute predictors (one for each attribute), and a content predictor and significantly outperforms five state-of-the-art methods.
- 11
- 11
- 11
- 11
- TargetSA: adaptive simulated annealing for target-specific drug design11
A novel framework that leverages adaptive simulated annealing (SA) for target-specific molecular generation and multi-constraint optimization and introduces a reversible sampling strategy to re-accept currently suboptimal solutions, greatly enhancing the generation quality.
- Evaluating the impact of test-trace-isolate for COVID-19 management and alternative strategies11
It is highlighted that, to reduce the overall transmission, it is important to limit contacts regardless of their types in addition to identifying infected individuals through contact tracing, given the substantial proportion of asymptomatic and pre-symptomatic SARS-CoV-2 transmission.
- Neural network with absent minority class samples and boundary shifting for imbalanced data classification11
An improved Metropolis Hasting (IMH) algorithm is developed to sample absent minority class samples by collecting samples rejected by the majority class approximation process, and this method demonstrates the effectiveness of sampling absent minorityclass samples for neural networks solving the imbalanced data problem.
- 11
- An automatic evaluation metric for Ancient-Modern Chinese translation11
This paper proposed an automatic evaluation metric for Ancient-Modern Chinese Translation called DTE (Dual-based Translation Evaluation), which can be used to evaluate one-to-many alignment in the absence of multiple references and designed a new word segmentation method to improve the translation of proper nouns without increasing the size of the model vocabulary.
- GSSA: Pay attention to graph feature importance for GCN via statistical self-attention11
This work attempts to augment convolutional layers in GCNs with statistical attention-based feature importance by modeling the latent interactions of features, which is complementary to the standard GCNs and only needs simple calculations with statistics rather than heavy trainings.
- Classification model of restricted Boltzmann machine based on reconstruction error11
This paper demonstrates that the reconstruction errors of the ClassRBM are larger than those of the conventional RBM because of the label information, and proposes a classification model of the restricted Boltzmann machine based on these reconstruction errors.
- 10
- Enhancing Diffusion Model Stability for Image Restoration via Gradient Management10
Stabilized Progressive Gradient Diffusion (SPGD) is proposed, a novel gradient management technique that integrates two synergistic components: a progressive likelihood warm-up strategy to mitigate gradient conflicts and adaptive directional momentum smoothing to reduce fluctuations in the likelihood gradient.
- EBAD-Gaussian: Event-driven Bundle Adjusted Deblur Gaussian Splatting10
This work proposes Event-driven Bundle Adjusted Deblur Gaussian Splatting (EBAD-Gaussian), which reconstructs sharp 3D Gaussians from event streams and severely blurred images, and optimize the latent sharp images at intermediate exposure times based on the event-based double integral (EDI) prior.
- Learning with Noisy Correspondence10
The experimental results show the necessity of the explicit solution to the NC faced by not only the standard model training paradigm but also the pre-training and fine-tuning paradigms.
- Deep Lifelong Cross-Modal Hashing10
This paper proposes a novel deep lifelong cross-modal hashing to achieve lifelong hashing retrieval instead of re-training hash function repeatedly when new data arrive, and introduces enhanced-semantic similarity to supervise hash learning and proves that the similarity improves performance with detailed analysis.
- FLSGD: free local SGD with parallel synchronization10
This work proposes a novel method, free local stochastic gradient descent (FLSGD) with parallel synchronization, to eliminate the waiting and communication overhead of distributed DNN training and is the first work focusing on decreasing both distributed training load balancing and Communication overhead.
- Multistage semantic‐aware image inpainting with stacked generator networks10
A stacked generator networks assisted by patch discriminator for image inpainting by multistage, which mainly consists of three‐layer stacked encoder‐decoder architecture, which could fuse different level feature information and achieve image in Painting via a coarse‐to‐fine hierarchical representation.
- 10
- Assessing the effectiveness of test-trace-isolate interventions using a multi-layered temporal network9
A model-based evaluation of Test-Trace-Isolate effectiveness highlights the challenges of relying on TTIs to contain an outbreak of a novel pathogen with characteristics similar to SARS-CoV-2 and estimates that the estimated effectiveness of TTI depends on the way contact patterns are modeled, supporting the relevance of obtaining comprehensive data on human social interactions to improve preparedness.
- Robust Contrastive Multi-view Kernel Clustering9
A Robust Contrastive multi-view kernel Learning approach (R-CMK) against false negative pairs that partitions negative pairs into different intervals based on distance or similarity, and for false negative pairs, reverses their optimization gradient.
- Co-Matching: Towards Human-Machine Collaborative Legal Case Matching9
This work proposes a collaborative matching framework called Co-Matching, which encourages both the machine and the legal practitioner to participate in the matching process, integrating tacit knowledge, and introduces a method called ProtoEM to estimate human decision uncertainty, facilitating the probabilistic combination.
- Effective and Efficient Adversarial Detection for Vision-Language Models via A Single Vector9
A novel and effective iN-time Embedding-based AdveRSarial Image DEtection (NEARSIDE) method, which exploits a single vector that distilled from the hidden states of VLMs, which is called the attacking direction, to achieve the detection of adversarial images against benign ones in the input.
- 9
- 9
- 9
- 9
- Multi-view Image Generation by Cycle CVAE-GAN Networks9
A VAE and GAN framework with a dual-input U-Net generator and a patch discriminator are proposed to generate high-resolution and realistic target images and it is demonstrated that the proposed method achieves better results than the state-of-the-art models.
- Unsupervised Feature Learning for Heart Sounds Classification Using Autoencoder9
The experimental results demonstrate that the features from the autoencoder are suitable for heart sound classification, and convert the time-domain signals of the cardiac cycle into spectrograms and apply principal component analysis whitening to reduce the dimensionality of the spectrogram.
- Classification of Calligraphy Style Based on Convolutional Neural Network9
The experimental results prove that the proposed method can not only well identify the style of different calligraphers, but also have good performance in the classification of the font format of soft pen calligraphy and hard pencalligraphy.
- Finding a good initial configuration of parameters for restricted Boltzmann machine pre-training9
Principal component analysis is used to capture the principal component directions of the training data to retain a strong learning ability and significantly accelerates the learning speed of restricted Boltzmann machines.
- 8
- Near-Optimal Resilient Aggregation Rules for Distributed Learning Using 1-Center and 1-Mean Clustering with Outliers8
The analysis show that constant approximations to the 1-center and 1-mean clustering problems with outliers provide near-optimal resilient aggregators for metric-based criteria, which have been proven to be crucial in the homogeneous and heterogeneous cases respectively.
- 8
- 8
- 8
- 8
- Cross-correlation conditional restricted Boltzmann machines for modeling motion style8
This paper presents cross-correlation as a new representation for modeling temporal dependencies by introducing the Pearson correlation coefficient and proposes an approach to enhance the discrimination of the CRBM for various human motions by incorporating cross-Correlation and temporal dependency features.
- Adaptive image zoom-in with bounding box transformation for UAV object detection7
The proposed ZoomDet is architecture-independent and can be applied to an arbitrary object detection architecture, and offers more than 8.4 absolute gain of mAP with a Faster R-CNN model, with only about 3 ms additional latency on the SeaDronesSee dataset.
- Aligning Information Capacity Between Vision and Language via Dense-to-Sparse Feature Distillation for Image-Text Matching7
This work argues that the information capacity of embeddings is crucial and proposes Dense-to-Sparse Feature Distilled Visual Semantic Embedding (D2S-VSE), which enhances the information capacity of sparse text by leveraging dense text distillation.
- Neural Boneprint: Person Identification from Bones Using Generative Contrastive Deep Learning7
This work exploits the thoracic skeletal data including chest radiographs (CXRs) and computed tomography (CT) images enhanced by the volume rendering technique (VRT) as an example to explore the availability of the neural boneprint.
- An Effective Augmented Lagrangian Method for Fine-Grained Multi-View Optimization7
The proposed Augmented Lagrangian MethOd for fiNe-graineD (ALMOND) multi-view optimization scrutinizes the interplay among multiple views at the granularity of individual samples, thereby fostering the enhanced preservation of local structural coherence.
- Abstract Rule Learning for Paraphrase Generation7
In RULER, a novel approach that performs abstract rule learning for paraphrasing by proposing a rule generalizability metric to guide the model to generate rules underlying the paraphrase, and leveraging neural networks to generate paraphrases by refining the sentences transformed by the learned rules.
- 7
- Multimodal image-to-image translation between domains with high internal variability7
This work recast the training procedure as modeling distinct distributions which are observed sequentially, for example, when different classes are encountered over time, and proposes a novel generator regulating GAN (GR-GAN), which encourages the discriminator to teach the generator more effectively when it remembers more of the previously learned distributions.
- U-Sleep: A Deep Neural Network for Automated Detection of Sleep Arousals Using Multiple PSGs7
A physiological time series segmentation network U-Sleep based on deep learning is proposed to analyze these sleep disorders, which uses the sequence to sequence input-output mode to map multiple complete original polysomnograms to a single tag sequence.
- 7
- A New Sparse Restricted Boltzmann Machine7
Although existing sparse restricted Boltzmann machine (SRBM) can make some hidden units activated, the major disadvantage is that the sparseness of data distribution is usually overlooked and the r...
- 6
- Genetic Auto-prompt Learning for Pre-trained Code Intelligence Language Models6
GenAP is the first work to automatically design prompts for code intelligence PLMs, which utilizes an elaborate genetic algorithm to automatically design prompts and outperforms all other methods across all three tasks.
- See the Unseen: Better Context-Consistent Knowledge-Editing by Noises6
It is empirically found that the effects of different contexts upon LLMs in recalling the same knowledge follow a Gaussian-like distribution, which can make LLMs see the unseen contexts where the edited knowledge will be applied, therefore improving the editing generalization.
- 6
- 6
- Automatic 3D Skeleton-based Dynamic Hand Gesture Recognition Using Multi-Layer Convolutional LSTM6
A Deep Convolutional LSTM (DConvLSTM) model is presented to learn more discriminative spatiotemporal features from skeleton data implicitly and can achieve comparable recognition accuracy while maintaining small models and short inference time.
- Combination of certainty and uncertainty: Using FusionGAN to create abstract paintings6
A novel FusionGAN system to automate the generation of abstract paintings that combines the factors of certainty and uncertainty and has been recognized by art institutions with some results displayed on their websites.
- Generating Chinese Poetry from Images via Concrete and Abstract Information6
This paper proposes an infilling-based Chinese poetry generation model which can infill the Concrete keywords into each line of poems in an explicit way, and an abstract information embedding to integrate the Abstract information into generated poems.
- Robust geodesic based outlier detection for class imbalance problem6
A robust geodesic-based outlier detection algorithm which simultaneously considers both global disconnectivity score and local real degree as measures of outlierness and can identify local outliers with higher overall connectivity but in a smaller cluster with fewer points.
- 6
- 6
- Exaggerated portrait caricatures synthesis6
A novel DeepExaggeration system to generate exaggerated portrait caricatures that successfully relieves the difficulties from using convolution neural networks for geometric structural features extraction and is the first to create the Magic Mirror Caricatures dataset.
- Automatic detection of boundary points based on local geometrical measures6
This paper presents an angle and density-based data preprocessing method that can be used to simultaneously identify outliers and boundary points (called uniformly boundary points) and demonstrates the effectiveness and efficiency of this method.
- Computer-Generated Abstract Paintings Oriented by the Color Composition of Images6
An efficient k-means-based method to segment the colors of an original picture to analyze the composition ratio of the color information and calculate individual color areas that are associated with their sizes is presented.
- From structure to synergy: A survey of vision-language perception paradigm evolution in multimodal large language models5
This study formalizes MLLM perception as an intrinsic, unified vision-language capability analogous to human innate perception and introduces a five-stage taxonomy tracing the paradigm evolution of MLLM perception and survey representative methods and milestones at each phase.
- E-3SFC: Communication-Efficient Federated Learning With Double-Way Features Synthesizing5
A novel gradient compression method termed 3SFC is introduced, which utilizes the model itself as a decompressor, leveraging training priors such as model weights and objective functions, and which can significantly enhance communication efficiency in FL without compromising model performance.
- PALA: Class-imbalanced Graph Domain Adaptation via Prototype-anchored Learning and Alignment5
This paper incorporates pointwise node mutual information into the graph encoder to capture high-order topological proximity and learn generalized node representations and introduces a weighted prototype contrastive adaptation strategy that aligns target pseudo-labels with source prototypes to handle class imbalance during adaptation.
- 5
- Adaptive Instance-wise Multi-view Clustering5
This work designs a collaborative framework with the augmented Lagrangian method to refine all subtasks towards optimal solutions iteratively and confirms the significant enhancement in clustering accuracy achieved by the method.
- 5
- Face deblurring based on regularized structure and enhanced texture information5
This work first incorporates the face parsing network with fine-tuning to obtain more accurate face structure, and presents the feature adaptive denormalization (FAD) to regularize the facial structure as a condition of auxiliary to generate more harmonious and undistorted face structure.
- 5
- 5
- PSO-PS:Parameter Synchronization with Particle Swarm Optimization for Distributed Training of Deep Neural Networks5
A new algorithm of integrating Particle Swarm Optimization (PSO) into the distributed training process of DNNs to automatically compute new parameters to reduce the communication and synchronization overhead of synchronous methods.
- TL-GAN: Generative Adversarial Networks with Transfer Learning for Mode Collapse (S)5
A generative adversarial network with transfer learning for mode collapse called TL-GAN is proposed, which can guarantee the diversity of samples and enhances the learning capability of the generator, which learn further about the distribution of real data.
- A Semi-supervised manifold alignment algorithm and an evaluation method based on local structure preservation5
It is shown that the proposed semi-supervised alignment method preserves the local geometry of the original dataset well, and it is found that better structural stability gives higher alignment accuracy, but the converse is not true.
- A non-negative representation learning algorithm for selecting neighbors5
A non-negative representation learning algorithm is proposed to select the neighbors of a data point, so that they are almost always lying in the tangent space of the manifold at the given query point and can preserve the local manifold structure well.
- A globally convergent learning algorithm for PCA neural networks5
A new PCA learning algorithm based on cascade recursive least square (CRLS) neural network is proposed that can guarantee the network weight vector converges to an eigenvector associated with the largest eigenvalue of the input covariance matrix globally.
- 4
- GDM-BC: Non-invasive body composition dataset for intelligent prediction of Gestational Diabetes Mellitus4
This paper introduces a large-scale non-invasive body composition dataset, called GDM-BC, for intelligent risk prediction of GDM, which contains a cohort of 39,438 pregnant women, of whom 7777 (19.7%) were subsequently diagnosed with GDM.
- Deep Disease Label-guided Graph Convolutional Network for Medical Report Generation4
A novel deep label-guided graph convolutional network for medical report generation which utilizes disease label to guide to extract pathological information from medical images and can be easily applied and extended to medical image report generation with different modalities.
- Learning Dynamic Similarity by Bidirectional Hierarchical Sliding Semantic Probe for Efficient Text Video Retrieval4
This paper proposes a novel method called Bidirectional Hierarchical Sliding Semantic Probe (BiHSSP), which calculates dynamic similarity between videos and texts with O(n) complexity during retrieval, and introduces a hierarchical semantic probe module that learns semantic probes at different scales for both video and text features.
- 4
- 4
- Dishonesty in Helpful and Harmless Alignment4
It is theoretically shown that the dishonesty can in-turn decrease the alignment performances and augment reward-seeking alignment with representation regularization, and it is theoretically shown that the dishonesty can in-turn decrease the alignment performances and augment reward-seeking alignment with representation regularization.
- 4
- DREditor: An Time-efficient Approach for Building a Domain-specific Dense Retrieval Model4
DREditor is developed to edit the matching rule of an off-the-shelf dense retrieval model to suit a specific domain by directly calibrating the output embeddings of the model using an efficient and effective linear mapping.
- 4
- With a Little Help from Language: Semantic Enhanced Visual Prototype Framework for Few-Shot Learning4
SEVpro refines prototype learning from the pre-training stage and serves as a versatile plug-and-play framework for all prototype-based FSL methods, enhancing prototype discriminability by transforming semantic embeddings into the visual space, aiding in separating categories with similar visual features.
- Differentiable Genetic Programming for High-dimensional Symbolic Regression4
A differentiable approach named DGP is proposed to construct GP trees towards high-dimensional SR for the first time and can facilitate SR being a powerful alternative to interpretable ML for a broader range of real-world problems.
- Highly effective end-to-end single-to-multichannel feature fusion and ensemble classification to decode emotional secretes from small-scale spontaneous facial micro-expressions4
Two highly effective, lightweight, and generalized single-channel DLRRF-MER, and multi-channel DLH-3C-FUSION fusion models inspired by deep dense convolutional models and texture-based feature descriptors are proposed to recognize ME from apex frame.
- 4
- CoupGAN: Chinese couplet generation via encoder–decoder model and adversarial training under global control4
This paper introduces a discriminative model to evaluate the global-level quality of generated sentences, named CoupGAN, and demonstrates that the proposed model is superior to existing methods in various evaluation metrics.
- 4
- Multiple partitions alignment via spectral rotation4
This paper proposes to merge multi-view information in partition level instead of the raw feature space where the data points lie, and formulated as a joint learning framework, i.e., with the input data matrix, the model directly outputs the final discrete clustering result.
- Personalized Federated Learning with Hidden Information on Personalized Prior4
Bayesian learning is employed to model PFL by assuming a prior in the scaled exponential family, and therefore pFedBreD, a framework to solve the problem the authors model using Bregman divergence regularization significantly outcompetes other PFL algorithms on multiple public benchmarks.
- 4
- A conditional classification recurrent RBM for improved series mid-term forecasting4
A new mid-term forecasting with error suppression based on restricted Boltzmann machine(RBM) is proposed in this paper and can suppress the error amplification by replacing the previous inputs with their features, which are retrieved by a deep belief network (DBN).
- 4
- PWGAN4
A novel Wasserstein Generative Adversarial Networks with perceptual loss function (PWGAN) is proposed in this paper, and experimental results show that the images generated by PWGAN have achieved better quality in visual effect and stability than state-of-the-art approaches.
- 4
- 4
- A surgical simulation system for predicting facial soft tissue deformation4
A surgical simulation system to predict facial soft tissue changes caused by the movement of bone segments duringCMF surgery and is applicable to computer-assisted CMF surgery is presented.
- 4
- Facial micro-expression recognition from videos through domain adaptation and multi-modal spatio-temporal feature ensemble3
This study proposes the deep model DITRAGS-STNet and the FHOG-DITRAGS-STNet comprising the dynamic texture handcrafted feature descriptor HOG-TOP and the optical flow feature descriptor Bi-WOOF, and the smaller size DITRAGS-STNet deep model to classify ME videos.
- Style Quantization for Data-Efficient GAN Training3
SQ-GAN, a novel approach that enhances CR by introducing a style space quantization scheme, transforms the sparse, continuous input latent space into a compact, structured discrete proxy space, allowing each element to correspond to a specific real data point, thereby improving CR performance.
- 3
- Target-Guided Bayesian Flow Networks for Quantitatively Constrained CAD Generation3
This work proposes a novel framework for quantitatively constrained CAD generation, termed Target-Guided Bayesian Flow Network (TGBFN), which handles the multi-modality of CAD sequences in a unified continuous and differentiable parameter space rather than in the discrete data space.
- An Empirical Study of Parameter Efficient Fine-tuning on Vision-Language Pre-train Model3
For downstream fine-tuning consistent with pre-training, data size no longer affects the performance, while the influence of fine-tunable parameter size is not monotonous, which could guide the choice of training strategy for various PEFTs.
- Estimating the time-varying effective reproduction number via Cycle Threshold-based Transformer3
The Ct-based deep learning method can improve the real-time estimates of Rt, enabling more easily adapted to the track of the newly emerged epidemic, and the supervised learning of Ct-Transformer outperforms the traditional incidence-based statistic and Ct-based Rt estimating methods.
- 3
- MS$^3$D: A RG Flow-Based Regularization for GAN Training with Limited Data3
The challenge of training GANs with limited data is investigated, and a novel regularization method based on the idea of renormalization group (RG) in physics is proposed, which constrains the gradient field to have a consistent pattern across different scales, thereby fostering a more redundant and robust system.
- 3
- Preserving Local and Global Information: An Effective Metric-based Subspace Clustering3
This paper proposes a novel subspace clustering method by holistically utilizing the pairwise similarity and graph geometric structure, which is able to depict the global structure of data and facilitate the downstream subspace learning task by searching for a smooth representation of the original data.
- 3
- Semantic-Pixel Associative Information Improving Loop Closure Detection and Experience Map Building for Efficient Visual Representation3
This paper calculates the similarity between current and previous scenes in LCD based on the pixel information computed by the sum of absolute differences (SAD) and the semantic information extracted by the YOLOv2 network and builds an enhanced experience map with object-level information.
- 3
- Multi-scale progressive blind face deblurring3
A multi-scale progressive face-deblurring generative adversarial network that requires no facial priors to generate more realistic multi- scale deblurring results by one feed-forward process is proposed, which achieves better accuracy and visual quality against state-of-the-art methods.
- 3
- 3
- Convergence Analysis for Oja+ MCA Learning Algorithm3
In this paper, the fixed points of the DDT system are determined and an invariant set is obtained and the convergence of the Oja+’s MCA algorithm via a corresponding deterministic discrete-time dynamical system is proven.
- Task-specific pre-training for molecular property prediction2
TasProp is proposed, a task-specific pre-training strategy for molecular property prediction, particularly for the scenarios with small labeled datasets, and an interactive web resource to facilitate model exploration and application, allowing users to easily predict the properties of input molecules online.
- Multimodal deep learning with routine clinical data for recurrence risk stratification in HR+/HER2- early breast cancer2
This study provides a clinically practical, AI-driven tool for precise risk stratification in HR+/HER2− EBC patients, facilitating individualized treatment and surveillance decisions without reliance on costly multi-omics data.
- Dual Prototype-Enhanced Contrastive Framework for Class-Imbalanced Graph Domain Adaptation2
This work proposes a dual-branch prototype-enhanced contrastive framework for graph domain adaptation under a class-imbalanced scenario, and introduces a dual-branch graph encoder to capture both local and global information, generating class-specific prototypes from a distilled anchor set.
- 2
- FAS: Fast ANN-SNN Conversion for Spiking Large Language Models2
A novel Fast ANN-SNN conversion strategy (FAS) that transforms LLMs into spiking LLMs in two stages that employs a full-parameter fine-tuning of pre-trained models, so it does not need any direct training from scratch.
- 2
- 2
- SODAS: Second-order optimization differential architecture search for diabetic retinopathy prediction2
This paper proposes second-order optimization differential architecture search (SODAS) for predicting diabetic retinopathy and integrates Gumbel-Softmax sampling into neural architecture search to encompass all the operation information during normalization, with the aims of reducing gradient information loss and improving accuracy.
- 2
- Dual Prompt Learning for Adapting Vision-Language Models to Downstream Image-Text Retrieval2
Dual prompt Learning with Joint Category-Attribute Reweighting (DCAR), a novel dual-prompt learning framework to achieve precise image-text matching that jointly optimizes attribute and category features to enhance fine-grained representation learning.
- 2
- 2
- 2
- Precision Neural Network Quantization via Learnable Adaptive Modules2
This paper utilizes the Power Of Square root of Two (POST) as the basis for exponential quantization, effectively handling the bell-shaped distribution of neural network weights across various bit-widths while maintaining computational efficiency through a Look-Up Table method (LUT).
- 2
- BOB-YOLO: Balancing Optimization Binarized YOLO via Module-Wise Latency2
A binary neural network for object detection called BOB-YOLO is proposed to achieve a balanced performance in terms of computational speed, model size, and detection accuracy, and the Roofline Model is introduced to address these limitations by considering both computational complexity and parameter compression, ensuring high computational intensity.
- 2
- 2
- 2
- Knowing-how & Knowing-that: A New Task for Machine Comprehension of User Manuals2
This work introduces the Knowing-how&Knowing-that task that requires the model to answer factoid-style, procedure- style, and inconsistent questions about user manuals, and resolves this task by jointly representing the steps and facts in a graph TARA, which supports a unified inference of various questions.
- Shunting at Arbitrary Feature Levels via Spatial Disentanglement: Toward Selective Image Translation2
- Differentiable Search of Accurate and Robust Architectures2
The proposed DSARA to automatically search for the neural architectures that are accurate and robust after adversarial training improves the accuracy and the robustness upper bound of the searched architectures by carefully designing the placement of the cells and the proportional relationship of the filter numbers.
- Effective DemeapexNet: Revealing Spontaneous Facial Micro-Expressions2
The experimental results surpass the effectiveness of the state- of-the-art methods and express the higher model generalization, which subsequently can expedite the applications such as lie catching, homeland securities, criminal detections, business deal negotiations, and clinical diagnosis through psychoanalysis.
- DeFTA: A Plug-and-Play Decentralized Replacement for FedAvg2
Decentralized Federated Trusted Averaging (DeFTA), a decentralized FL framework that serves as a plug-and-play replacement for FedAvg, instantly bringing better security, scalability, and fault-tolerance to the federated learning process after installation.
- 2
- 2
- Method to Improve the Performance of Restricted Boltzmann Machines2
A method to train RBM by adding reconstruction error to the cost function is put forward and the experimental results show that the proposed method can be effective.
- Extracting Generic Features of Artistic Style via Deep Convolutional Neural Network2
This paper proposes a method to extract generic style features from a set of fine paintings using deep convolutional neural network, and obtains feature representation from these fine paintings, and selects generic representation from obtained representation.
- 2
- 1
- Mixture of Meta-Policies for Cross-Environment Meta-Reinforcement Learning1
Experiments across diverse MuJoCo agents show that MoMP significantly outperforms strong meta-RL baselines, highlighting its effectiveness in cross-environment generalization and efficient meta-policy reuse.
- CMGL: Confidence-guided Multi-omics Graph Learning for Cancer Subtype Classification1
CMGL is proposed, which produces a separate reliability estimate before fusion and uses consensus patient neighborhoods for graph classification and performs classification on an independently specified consistency graph.
- 1
- HyperNAS: Enhancing Architecture Representation for NAS Predictor via Hypernetwork1
This paper proposes HyperNAS, a novel neural predictor paradigm for enhancing architecture representation learning that combines a global encoding scheme and a shared hypernetwork, and develops a dynamic adaptive multi-task loss to facilitate personalized exploration on the Pareto front.
- 1
- Bridging the Gap between Learning and Inference for Diffusion-Based Molecule Generation1
DiffGap is introduced, a diffusion-based framework that integrates adaptive sampling and pseudo-molecule estimation to bridge the gap between training objectives and inference dynamics in 3D molecule generation, offering a robust computational toolkit for accelerating structure-based therapeutic discovery.
- 1
- VCL Challenges 2023 at ICCV 2023 Technical Report: Bi-level Adaptation Method for Test-time Adaptive Object Detection1
This report outlines the team's participation in VCL Challenges B Continual Test_time Adaptation, focusing on the technical details of the approach, and achieves a remarkable 38.3% mAP on the target domain of the test set within VCL challenges B.
- 1
- Lateral Interactions Spiking Actor Network for Reinforcement Learning1
Lateral Interactions Spiking Actor Network (LISAN) is proposed to improve decision-making in reinforcement learning tasks with high performance and incorporate soft reset mechanism to enhance model’s functionality recognizing the significance of residual potentials in preserving valuable information within biological neurons.
- Research of Flight Safety and Technical Evaluation1
A flight technology evaluation method based on flight parameter data is discussed, and it is shown that the BP neural network verification set is more accurate than the decision tree model.
- 1
- 1
- 1
- Correction to: FLSGD: free local SGD with parallel synchronization1
This paper presents a probabilistic procedure for estimating the number of nodes in a distributed supercomputing system using a simple, scalable, and scalable approaches called "cells".
- Sparse DARTS with Various Recovery Algorithms1
Sufficient theoretical analysis and experimental exploration reveal that the different compression algorithms show different characteristics on the sparse DARTS, and all compression algorithms are available to utilized according to different environments and requirements.
- Cluster-based Contrastive Disentangling for Generalized Zero-Shot Learning1
A Cluster-based Contrastive Disentangling (CCD) method to improve GZSL by alleviating the semantic gap and domain shift problems and introduces contrastive learning on semantic-matched and class-unique variables to learn high intra-set and intra-class similarity.
- 1
- 1
- 1
- A Data Augmentation Model Based on Variational Approach1
A new data augmentation model that can extract the deformation features between the given deformation image and the original image is proposed, which proves that the new sample is effective.
- Tree Factored Conditional Restricted Boltzmann Machines for Mixed Motion Style1
The proposed tree-based FCRBM model can make the newly generated data have a variety of sports styles and achieve a smooth transition between the styles using little or even no labeled data.
- 1
- 1
- 1
- TFTA: cross-topic rumor detection with time-aware attention–
This work proposes a two-stage cross-topic rumor detection model that significantly outperforms state-of-the-art methods in both cross-topic rumor detection and early detection tasks, and introduces a time-aware graph attention mechanism with adversarial training to further optimize feature fusion.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- GPS: Distilling Compact Memories via Grid-based Patch Sampling for Efficient Online Class-Incremental Learning–
Grid-based Patch Sampling (GRS) is introduced, a lightweight and effective strategy for distilling informative memory samples without relying on a trainable model, leading to 3%-4% improvements in average end accuracy under memory-constrained settings, with limited computational overhead.
- –
- –
- –
- –
- –
- –
- –
- DONIS: Importance Sampling for Training Physics-Informed DeepONet–
This work presents an enhanced PI-DeepONet framework, that applies importance sampling to both of DeepONet inputs (i.e., the functions and the collocation points) to alleviate training challenges and showcases accelerated convergence and improved accuracy across various complex applications.
- –
- –
- –
- Learning Locally, Revising Globally: Global Reviser for Federated Learning with Noisy Labels–
This paper proposes a novel method termed Federated Global Reviser (\method), a straightforward yet effective method comprising three modules that collaboratively rectify noisy labels and regularize local training that improves the label-noise robustness of FL in a self-contained manner.
- –
- –
- Analysis and Applications of Deep Learning with Finite Samples in Full Life-Cycle Intelligence of Nuclear Power Generation–
The present study centers on the intricate and distinctive industrial scenarios of Nuclear Power Generation (NPG), meticulously scrutinizing the application of DL techniques under the constraints of finite data samples and introduces a fresh perspective toward the secure and efficient advancement and exploitation of this advanced energy source.
- Graph-level Anomaly Detection via Deep Metric Learning–
This work re-consider GAD tasks from another perspective and proposes a novel end-to-end method to spot graph anomalies, which can effectively detect graph-level anomalies and performs competitively against other state-of-the-art methods.
- Attribute Localization and Revision Network for Zero-Shot Learning–
It is found that the choice between local and global features is not a zero-sum game, global features can also contribute to the understanding of attributes, and a novel module is incorporated to fuse global and local representations.
- Draft, Command, and Edit: Controllable Text Editing in E-Commerce–
A new draft-command-edit manner in description generation is explored, leading to the proposed new task—controllable text editing in E-commerce and a data augmentation method is designed to remedy the low resource challenge in this task.
- The optimal buyback promotion strategy considering consumers' multichannel psychological acquisition–
- –
- HPSGD: Hierarchical Parallel SGD with Stale Gradients Featuring–
A novel Hierarchical Parallel SGD strategy is proposed based on the observation that the data synchronization phase can be paralleled with the local training phase, which substantially boosts the distributed DNN training, reduces the disturbance of the stale gradients and achieves better accuracy in given fixed wall-time.
- HLSGD Hierarchical Local SGD With Stale Gradients Featuring.–
A novel Hierarchical Parallel SGD strategy is proposed based on the observation that the data synchronization phase can be paralleled with the local training phase, which substantially boosts the distributed DNN training, reduces the disturbance of the stale gradients and achieves better accuracy in given fixed wall-time.
- –
- –
- –
- Novel Online Dimensionality Reduction Method with Improved Topology Representing and Radial Basis Function Networks–
A novel method for online dimensionality reduction that integrates the improved Topology Representing Network and Radial Basis Function Network is proposed that can find meaningful low-dimensional feature structures embedded in high-dimensional original data space, process nonlinear embedded manifolds, and map the new data online.
- –
- –
- –
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-12. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.