Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Arne LeijonSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works327 from public data
- Dual Graph Convolutional Networks for Aspect-based Sentiment Analysis472
A dual graph convolutional networks (DualGCN) model that considers the complementarity of syntax structures and semantic correlations simultaneously and proposes orthogonal and differential regularizers to capture semantic correlations between words precisely by constraining attention scores in the SemGCN module.
- The Devil is in the Channels: Mutual-Channel Loss for Fine-Grained Image Classification401
Experimental results show the superiority of the MC-Loss when implemented on top of common base networks can achieve state-of-the-art performance on all four fine-grained categorization datasets, and ablative studies further demonstrate the supremacy of the loss when compared with other recently proposed general-purpose losses for visual classification.
- Fine-Grained Visual Classification via Progressive Multi-granularity Training of Jigsaw Patches389
This work proposes a novel framework for fine-grained visual classification with a progressive training strategy that effectively fuses features from different granularities, and a random jigsaw patch generator that encourages the network to learn features at specificgranularities.
- 332
- Prediction of short-term PV power output and uncertainty analysis281
The results imply that the two-stage model proposed in the paper outperforms conventional forecast methods in terms of prediction of short-term PV power outputs and associated uncertainties.
- 269
- 238
- Deep metric learning for few-shot image classification: A Review of recent developments237
This paper provides an up-to-date review of deep metric learning methods for few-shot image classification from 2018 to 2022 and categorizes them into three groups according to three stages of metric learning, namely learning feature embeddings, learning class representations, and learning distance measures.
- BSNet: Bi-Similarity Network for Few-shot Fine-grained Image Classification214
A so-called Bi-Similarity Network (BSNet) that consists of a single embedding module and a bi-similarity module of two similarity measures of diverse characteristics that is enabled to learn more discriminative and less similarity-biased features from few shots of fine-grained images, such that the model generalization ability can be significantly improved.
- Your “Flamingo” is My “Bird”: Fine-Grained, or Not155
The key intuition that: coarse- level label prediction exacerbates fine-grained feature learning, yet fine-level feature betters the learning of coarse-level classifier is discovered, which enables the design of a very simple albeit surprisingly effective solution to a new problem.
- 146
- A concise review of recent few-shot meta-learning methods145
This short communication gives a concise review on recent representative methods in few-shot meta-learning, which are categorized into four branches according to their technical characteristics.
- The Role of Data Analysis in the Development of Intelligent Energy Networks145
More comprehensive data analysis methods are needed to handle the increasing amount of data and to mine more valuable information in intelligent energy networks.
- 138
- Duplex Contextual Relation Network For Polyp Segmentation136
It is argued that exploring contextual correlation from a holistic view of the whole dataset is essential and a Duplex Contextual Relation Network (DCRNet) is proposed to capture both within-image and cross-image contextual relations.
- 133
- DemoFusion: Democratising High-Resolution Image Generation With No $$$131
This paper aims to democratise high-resolution GenAl by advancing the frontier of high-resolution generation while remaining accessible to a broad audience, and demonstrates that existing Latent Diffusion Models (LDMs) possess untapped potential for higher-resolution image generation.
- 130
- SketchMate: Deep Hashing for Million-Scale Human Sketch Retrieval130
This work proposes a deep hashing framework for sketch retrieval that, for the first time, works on a multi-million scale human sketch dataset and shows that state-of-the-art hashing models specifically engineered for static images fail to perform well on temporal sketch data.
- Decorrelation of Neutral Vector Variables: Theory and Applications121
Two fundamental invertible transformations, namely, serial non linear transformation and parallel nonlinear transformation, are proposed to carry out the decorrelation, which shows that a highly negatively correlated neutral vector can be transformed to a set of mutually independent scalar variables with the same degrees of freedom.
- 117
- Efficient Face Super-Resolution via Wavelet-based Feature Enhancement Network115
A wavelet-based feature enhancement network is proposed, which mitigates feature distortion by losslessly decomposing the input feature into high and low-frequency components using the wavelet transform and processing them separately.
- 115
- Bayesian estimation of Dirichlet mixture model with variational inference114
The extended factorized approximation method is applied to introduce a single lower-bound to the variational objective function and an analytically tractable estimation solution is derived and the convergence of the proposed algorithm is theoretically guaranteed.
- Fine-Grained Age Estimation in the Wild With Attention LSTM Networks113
Experimental results illustrate the effectiveness and robustness of the proposed AL-ResNets or AL-RoR for age estimation in the wild, where it achieves better state-of-the-art performance than all other convolutional neural network methods on the Adience, MORPH Album 2, FG-NET and 15/16LAP datasets.
- 98
- 97
- 94
- Cross-modal subspace learning for fine-grained sketch-based image retrieval90
Through thorough examination of the experimental results, it is demonstrated that the subspace learning can effectively model the sketch-photo domain-gap, and a few key insights to drive future research are drawn.
- 78
- 77
- 74
- 71
- Domain Generalization via Frequency-domain-based Feature Disentanglement and Interaction71
By leveraging the frequency domain of an image, this paper uniquely work with two key observations: the high-frequency information of animage depicts object edge structure, which preserves high-level semantic information of the object is naturally consistent across different domains, and the low-frequency component retains object smooth structure, while this information is susceptible to domain shifts.
- Computer-aided diagnosis of colorectal polyps using linked color imaging colonoscopy to predict histology69
A computer-aided diagnosis (CAD) system based on linked color imaging (LCI) images to predict the histological results of polyps by analyzing the colors of the lesions could be a rapid and powerful decision-making tool for endoscopists.
- DNN Filter Bank Cepstral Coefficients for Spoofing Detection67
A new filter bank-based cepstral feature, deep neural network (DNN) filter bank cepStral coefficients, to distinguish between natural and spoofed speech, and the experimental results show that the Gaussian mixture model maximum-likelihood classifier trained by the new feature performs better than the state-of-the-art linear frequency triangle filter bankcepstrals-based classifier, especially on detecting unknown attacks.
- Progressive Co-Attention Network for Fine-Grained Visual Classification64
This paper calculates the channel-wise similarity by encouraging interaction between the feature channels within same-category image pairs to capture the common discriminative features and erase the prominent areas enhanced by the channel interaction to force the network to focus on other discriminating regions.
- Advanced Dropout: A Model-free Methodology for Bayesian Dropout Optimization63
The proposed advanced dropout technique applies a model-free and easily implemented distribution with parametric prior, and adaptively adjusts dropout rate, and is extended to uncertainty inference, network pruning, text classification, and regression.
- Semi-Heterogeneous Three-Way Joint Embedding Network for Sketch-Based Image Retrieval61
A novel semi-heterogeneous three-way joint embedding network (Semi3-Net) is proposed, which integrates three branches (a sketch branch, a natural image branch, and an edgemap branch) to learn more discriminative cross-domain feature representations for the SBIR task.
- 59
- Image-text dual neural network with decision strategy for small-sample image classification59
This work shows how to efficiently and effectively utilize semantic information of the annotations to improve the performance of small-sample classification by incorporating a decision strategy into the image-text dual neural network to further improve theperformance of the original model on few-shot datasets.
- 54
- OSLNet: Deep Small-Sample Classification With an Orthogonal Softmax Layer52
The proposed Orthogonal Softmax Layer (OSL), which makes the weight vectors in the classification layer remain orthogonal during both the training and test processes, has better performance than the methods used for comparison on four small-sample benchmark datasets, as well as its applicability to large-sample datasets.
- A Survey on Machine Learning‐Based Mobile Big Data Analysis: Challenges and Applications52
This paper attempts to identify the requirement and the development of machine learning-based mobile big data (MBD) analysis through discussing the insights of challenges in the mobile bigData, and reviews the state-of-the-art applications of data analysis in the area of MBD.
- A Cross-Domain Recommendation Model for Cyber-Physical Systems50
A novel cross-domain recommendation model, which not only learns the common rating pattern across domains with the flexibility in controlling the optimal level of sharing, but also learn the domain-specific rating patterns in each domain involving discriminative information propitious to performance improvement is proposed.
- 50
- Self-reconstruction network for fine-grained few-shot classification49
A self-reconstruction metric module for diversifying query features and a restrained cross-entropy loss for avoiding over-confident predictions are proposed to effectively alleviate overfitting in metric-based few-shot image classification.
- 47
- 46
- 45
- GINet: Graph Interaction Network for Scene Parsing45
This work explores how to incorporate the linguistic knowledge to promote context reasoning over image regions by proposing a Graph Interaction unit (GI unit) and a Semantic Context Loss (SC-loss).
- Adversarial Network Bottleneck Features for Noise Robust Speaker Verification45
Experimental results on the RSR2015 database show that the proposed AN bottleneck feature (AN-BN) dramatically outperforms the STSA-MMSE and DNN-SE based MFCCs for different noise types and signal-to-noise ratios and is able to improve the speaker verification performance under the clean condition.
- Learning Deep Features for DNA Methylation Data Analysis45
A deep neural network is designed, which composes of several stacked binary restricted Boltzmann machines, to learn the low-dimensional deep features of the DNA methylation data, which perform best in breast cancer data cluster analysis, compared with some state-of-the-art methods.
- VisualCloze: A Universal Image Generation Framework via Visual in-Context Learning44
Graph200K, a graph-structured dataset that establishes various interrelated tasks, enhancing task density and knowledge transfer, is introduced, and an intrinsic alignment between image infilling and in-context learning is uncovered, enabling us to leverage the strong generative priors of pre-trained infilling models without modifying their architectures.
- Feature selection for neutral vector in EEG signal classification44
Assuming the mDWT vector coefficients to be a neutral vector, the proposed parallel nonlinear transformation (PNT) framework is applied to transform them non-linearly into a set of independent scalar coefficients and feature selection strategy is proposed on the transformed feature domain.
- Dilated-Scale-Aware Category-Attention ConvNet for Multi-Class Object Counting42
This paper proposes a simple yet efficient counting network based on point-level annotations that first changes the traditional output channel from one to the number of categories to achieve multiclass counting, and designs a multi-mask structure to suppress harmful interaction among objects.
- 41
- 41
- Fine-Grained Instance-Level Sketch-Based Video Retrieval40
This work proposes a novel cross-modal retrieval problem of fine-grained instance-level sketch-based video retrieval (FG-SBVR), where a sketch sequence is used as a query to retrieve a specific target video instance.
- Analysis of Key Factors in Heat Demand Prediction with Neural Networks40
A model based on Elman neural networks was adopted, of which the results can help heat producers to optimize their production and thus mitigate costs and the introduction of wind speed and direct solar radiation has opposite impacts on the performance of ENN.
- 39
- 39
- 39
- An Erudite Fine-Grained Visual Classification Model38
This paper proposes an erudite FGVC model jointly trained by several different datasets to reduce negative transfer and boost positive transfer between different datasets, and proposes a meta-learning based dataset-agnostic spatial attention layer to take full advantage of the multi-dataset training data.
- 38
- LRID: A new metric of multi-class imbalance degree based on likelihood-ratio test38
A new metric based on the likelihood-ratio test, LRID, is proposed to provide a more reliable measurement of class-imbalance extent for multi-class data.
- NTIRE 2025 Challenge on Day and Night Raindrop Removal for Dual-Focused Images: Methods and Results37
This paper reviews the NTIRE 2025 Challenge on Day and Night Raindrop Removal for Dual-Focused Images to establish a new and powerful benchmark for the task of removing raindrops under varying lighting and focus conditions.
- 37
- GRAD-CAM Guided Channel-Spatial Attention Module for Fine-Grained Visual Classification35
A Grad-CAM guided channel-spatial attention module for the FGVC, which employs the gradients to supervise and constrain the attention weights by generating the coarse localization maps and demonstrates the superiority of the proposed method against the SOTA approaches.
- Construction of semantic bootstrapping models for relation extraction34
This paper introduces formalization for this class of bootstrapping models, which allows semantic constraint to guide learning iterations and use flexible bottom-up kernel to compare patterns and demonstrates performance superior to the state of the art.
- GPCA: A Probabilistic Framework for Gaussian Process Embedded Channel Attention32
The Gaussian process embedded channel attention (GPCA) module is proposed and can be efficiently implemented and integrated into the end-to-end training of the CNNs with a mathematically tractable solution.
- Shoe-Print Image Retrieval With Multi-Part Weighted CNN30
A novel multi-part weighted convolutional neural network (MP-CNN) for shoe-print image retrieval that significantly outperforms other fine-grained cross-domain methods on SPID dataset and obtains comparative results with the state-of-the-art shoe-prints retrieval methods on FID300 dataset.
- Data scheme-based wireless channel modeling method: motivation, principle and performance30
A channel-modeling method using PCA is proposed, whose principle is to utilize the features and structures extracted from the CIR data collected by measurements, and then model the wireless channel of the targeted measurement scenario.
- 29
- A VARIATIONAL BAYES BETA MIXTURE MODEL FOR FEATURE SELECTION IN DNA METHYLATION STUDIES29
It is demonstrated that VBBMM offers significant improvements in inference and feature selection in this type of data compared to an Expectation-Maximization (EM) algorithm, at a significantly reduced computational cost.
- Dual-Domain Modulation Network for Lightweight Image Super-Resolution28
This paper proposes a Dual-domain Modulation Network that integrates both wavelet and Fourier information for enhanced frequency modeling and achieves a comparable PSNR of SRFormer and MambaIR while with less than 50% and 60% of their FLOPs and achieving inference speeds 15.4× and 5.4× faster.
- NeRSP: Neural 3D Reconstruction for Reflective Objects with Sparse Polarized Images28
This paper jointly han-dles the challenges from sparse inputs and reflective surfaces by leveraging polarized images and derives photomet-ric and geometric cues from the polarimetric image formation model and multiview azimuth consistency, which jointly optimize the surface geometry modeled via implicit neural representation.
- On-the-Fly Category Discovery27
This paper designs a hash coding-based expandable recognition model and proposes a novel Sign-Magnitude dIsentangLEment (SMILE) architecture to alleviate the disturbance it brings, and experimental results demonstrate the superiority of SMILE against the baseline model and prior art.
- Text-Independent Speaker Identification Using the Histogram Transform Model27
Compared with the typical PDF estimation methods, such as Gaussian mixture model, promising improvements have been obtained by employing the HT-based model in SI and a smooth PDF of the super-MFCCs vectors is obtained.
- Instance-Level Coupled Subspace Learning for Fine-Grained Sketch-Based Image Retrieval27
This paper forms a cross-domain framework specifically designed for the task of FG-SBIR that simultaneously conducts instance-level retrieval and attribute prediction and proposes a joint view selection and attribute subspace learning algorithm to learn domain projection matrices for photo and sketch, respectively.
- 27
- Survey on Deep Face Restoration: From Non-blind to Blind and Beyond26
The prevalent factors responsible for real-world LQ images and degradation techniques used to synthesize LQ images are examined and degradation techniques used to synthesize LQ images are introduced.
- 26
- Non-Gaussian Statistical Models and Their Applications26
Evaluations with synthetic data and real life data demonstrate that the proposed Bayesian nonnegative matrix factor-ization method for bounded support data is efficient for source separation, missing data prediction, and collaborativefiltering problems.
- Modelling speech line spectral frequencies with dirichlet mixture models26
This dissertation studies several estimation methods for the non-Gaussian distributions and proposes a variational Bayesian Nonnegative Matrix Factorization (NMF) forbounded support data and a Variational Inference framework based method for approximating the parameter distribution.
- I-Max: Maximize the Resolution Potential of Pre-trained Rectified Flow Transformers with Projected Flow25
The I-Max framework is introduced and shown to enhance stability in resolution extrapolation and show that it can bring image detail emergence and artifact correction, confirming the practical value of tuning-free resolution extrapolation.
- Sea fog detection based on unsupervised domain adaptation25
This work proposes an unsupervised domain adaptation method to bridge the abundant labeled land fog data and the unlabeled sea fog data to realize the sea fog detection and achieves an accuracy of sea fog recognition up to 99.17%, which is nearly 3% higher than those vanilla methods.
- 25
- 25
- 25
- Task-aware Adaptive Learning for Cross-domain Few-shot Learning24
The Task-aware Adaptive Network (TA2-Net) is proposed, which is trained by reinforcement learning to adaptively estimate the optimal task-specific parameter configuration for each test task, and outperforms existing state-of-the-art methods.
- 24
- 23
- Supervised latent Dirichlet allocation with a mixture of sparse softmax23
This work embeds a mixture of softmax model into latent Dirichlet allocation model, and proposes a supervised topic model, supervised latent Diriclet allocation with a mixtureof softmax, and its improved version, supervised Latent Dirichlets allocation with an improved version of sparse softmax.
- Making a Bird AI Expert Work for You and Me22
This paper proposes to represent knowledge as highly discriminative visual regions that are expert-exclusive and instantiate it via a novel multi-stage learning framework and proposes a crude but benchmarkable metric TEMI to allow future efforts in this direction to be comparable to the authors' without the need of large-scale human studies.
- 22
- 22
- 22
- Dirichlet mixture modeling to estimate an empirical lower bound for LSF quantization22
The line spectral frequencies (LSFs) are commonly used for the linear predictive/autoregressive model in speech and audio coding and the probability density function (PDF)-optimized vector quantization (VQ) has been studied intensively for quantization of LSF parameters.
- 21
- 21
- Predictive Distribution of the Dirichlet Mixture Model by Local Variational Inference21
This paper proposes an algorithm to calculate the predictive distribution of the DMM with the local variational inference (LVI) method and considers the concave property of the multivariate inverse beta function to introduce an upper-bound to the true predictive distribution.
- PDF-optimized LSF vector quantization based on beta mixture models21
The LSF parameters are obtained from the TIMIT database and a practical VQ is designed, which outperforms the Gaussian mixture model based VQ with uncorrelated Gaussiancomponent (UGMVQ) by about 1-2 bits/vector.
- 21
- FourierSR: A Fourier Token-Based Plugin for Efficient Image Super-Resolution20
A Fourier token-based plugin called FourierSR is proposed to improve SR uniformly, which avoids the instability or inefficiency of existing token mix technologies when applied as plug-ins.
- Channel-Spatial Support-Query Cross-Attention for Fine-Grained Few-Shot Image Classification20
A new Channel-Spatial Cross-Attention Module (CSCAM) is proposed, which can effectively drive a model to extract discriminative fine-grained feature representations with only few shots, conveniently enabling effective improvement of state-of-the-art methods for few-shot fine-grained image classification.
- Toward Generalizable Forgery Detection and Reasoning19
Experiments across multiple generative models demonstrate that FakeReasoning not only achieves robust generalization but also outperforms state-of-the-art methods on both detection and reasoning tasks.
- Reduced loss aversion in value-based decision-making and edge-centric functional connectivity in patients with internet gaming disorder19
The reduced loss aversion in value-based decision making and their related edge-centric functional connectivity support that the IGD showed the same value- based decision-making deficit as the substance use and other behavioral addictive disorders.
- 19
- ReMarNet: Conjoint Relation and Margin Learning for Small-Sample Image Classification19
This paper proposes to enhance the discriminative power of features from a new perspective by introducing a novel neural network termed Relation-and-Margin learning Network (ReMarNet), which assembles two networks of different backbones so as to learn the features that can perform excellently in both of the aforementioned two classification mechanisms.
- Oscillation Resolution for Massive Cell Phone Traffic Data19
An algorithm called SOL (Stable, Oscillation, Leap periods) aimed at discovering and reducing oscillations in the collected logs of mobile subscribers is proposed and experimental results demonstrate the ability and effectiveness of SOL.
- 19
- 19
- 18
- 18
- Actor and Action Modular Network for Text-Based Video Segmentation18
A novel actor and action modular network that individually localizes the actor and its action in two separate modules is proposed that achieves the state-of-the-art performance for both single-frame segmentation and full video segmentation on A2D Sentences and J-HMDB Sentences datasets.
- User location prediction with energy efficiency model in the Long Term‐Evolution network18
This paper studies domain‐independent prediction algorithms and spatio‐temporal based prediction method using 20‐day‐long records in Long Term‐Evolution(LTE) network, which captures the mobility patterns of 3474 individuals and finds Markov provides the best tradeoff.
- Multi-label learning with prior knowledge for facial expression analysis18
A multi-label learning algorithm for FEA is proposed to solve the problem of depicting facial expressions more effectively and model FEA as a multi- label problem, which depicts all facial expressions with multiple continuous values and labels of predefined affective states.
- Dual-Prior Augmented Decoding Network for Long Tail Distribution in HOI Detection17
This paper proposes a dual-prior augmented decoding network by decomposing the HOI task into two sub-tasks: human-object pair detection and interaction recognition and leverages external knowledge to enhance the model's ability at a finer granularity.
- 17
- 17
- Line spectral frequencies modeling by a mixture of von Mises–Fisher distributions17
Experimental results show that the proposed VVQ outperforms the recently introduced Dirichlet mixture model-based VQ and the conventional Gaussian mixturemodel- based VQ, in terms of modeling performance and D-R relation.
- Mining activation force defined dependency patterns for relation extraction17
Experimental results demonstrate that the activation force based trigger word mining method plays an essential role in improving the performance, and the proposed method has good performance in the implementation of English Slot Filling 2013.
- 16
- ConMo: Controllable Motion Disentanglement and Recomposition for Zero-Shot Motion Transfer16
ConMo is introduced, a zero-shot framework that disentangle and recompose the motions of subjects and camera movements that unlocks a wide range of applications, including subject size and position editing, subject removal, semantic modifications, and camera motion simulation.
- 16
- 16
- 16
- 16
- 16
- 16
- 16
- Super-Resolution Information Enhancement for Crowd Counting15
This work proposes a more elegant method termed Multi-Scale Super-Resolution Module (MSSRM), which guides the network to estimate the lost details and enhances the detailed information in the feature space and proposes a Super-resolution Crowd Counting dataset (SR-Crowd).
- 14
- Animal-Bench: Benchmarking Multimodal Video Models for Animal-centric Video Understanding13
This work established an animal-centric benchmark, namely Animal-Bench, to allow for a comprehensive evaluation of model capabilities in real-world contexts, overcoming agent-bias in previous benchmarks.
- 13
- DS-UI: Dual-Supervised Mixture of Gaussian Mixture Models for Uncertainty Inference in Image Recognition13
Experimental results show the DS-UI outperforms the state-of-the-art UI methods in misclassification detection and finds statistically significant improvements in open-set out- of-domain/-distribution detection.
- Understanding Objects in Video: Object-Oriented Video Captioning via Structured Trajectory and Adversarial Learning13
The proposed STraNet, a video-based structured trajectory network via adversarial learning (STraNet) to effectively analyze the activities along the time domain and proactively capture the vision-language connections under small datasets, exhibits the ability of precisely describing the concurrent objects and their activities in details.
- Deep Zero-Shot Learning for Scene Sketch13
The augmented semantic vector is proposed to conduct domain alignment by fusing multi-modal semantic knowledge (e.g., cartoon image, natural image, text description), and adopt attention-based network for scene sketch feature learning.
- Super-dirichlet mixture models using differential line spectral frequencies for text-independent speaker identification13
A new text-independent speaker identification (SI) system that utilizes the line spectral frequencies (LSFs) as alternative feature set for capturing the speaker characteristics is proposed.
- 12
- Benchmarking Segmentation Models with Mask-Preserved Attribute Editing12
It is argued that local attributes have the same importance as global attributes, and should be considered in the robustness evaluation of segmentation models, and the sensitivity of models diverges across different variation types.
- 12
- CC-Loss: Channel Correlation Loss for Image Classification12
A channel correlation loss (CC-Loss) is proposed that is able to constrain the specific relations between classes and channels as well as maintain the intra-class and the inter-class separability.
- 12
- MedReasoner: Reinforcement Learning Drives Reasoning Grounding from Clinical Thought to Pixel-Level Precision11
This work defines Unified Medical Reasoning Grounding (UMRG) and introduces MedReasoner, a modular framework that distinctly separates reasoning from segmentation: an MLLM reasoner is optimized with reinforcement learning, while a frozen segmentation expert converts spatial prompts into masks, with alignment achieved through format and accuracy rewards.
- S2TD: A Tree-Structured Decoder for Image Paragraph Captioning11
A novel tree-structured visual paragraph decoder network, called Splitting to Tree Decoder (S2TD) is proposed to address this problem, to model the paragraph decoding process as a top-down binary tree expansion.
- Comparisons of Non-Gaussian Statistical Models in DNA Methylation Analysis11
Experimental results show that the non-Gaussian statistical model-based methods are superior to the conventional Gaussian distribution-based method and are meaningful tools for DNA methylation analysis.
- Human audio-visual consonant recognition analyzed with three bimodal integration models11
With A-V recordings, normal hearing people took recognition tests at different signal-to-noise ratios (SNR) and all the models agree qualitatively with the results that the benefit gained from the visual signal is larger at lower acoustic SNRs.
- PolGS: Polarimetric Gaussian Splatting for Fast Reflective Surface Reconstruction10
The proposed PolGS is a arimetric arimetric aussian aussian latting model allowing fast reflective surface reconstruction in 10 minutes, and takes the shortest time while comparable shape reconstruction accuracy is measured by Chamfer Distance in millimeters.
- Zero-Shot Audio Captioning Using Soft and Hard Prompts10
This work proposes a new zero-shot method for audio captioning, built on the contrastive language-audio pre-training (CLAP) model, which outperforms other zero-shot audio captioning approaches for in-domain scenarios and outperforms the compared methods for cross-domain scenarios.
- 10
- Category-Specific Prompts for Animal Action Recognition with Pretrained Vision-Language Models10
Experimental results demonstrate that the method outperforms five previous action recognition methods on the Animal Kingdom dataset and has shown best generalization ability on unseen animals.
- 10
- On von-mises fisher mixture model in text-independent speaker identification10
This paper analytically estimates the mixture model parameters in a fully Bayesian treatment by using variational inference and shows that the square root of MLSF has interesting directional characteristics implying that their distribution can be modeled by a mixture of von-Mises Fisher (vMF) distributions.
- Animal-CLIP: A Dual-Prompt Enhanced Vision-Language Model for Animal Action Recognition9
Animal-CLIP is introduced, a specialized and efficient animal action recognition framework built upon existing VLMs that outperforms six previous action recognition methods across three large-scale multi-species, multi-action datasets and exhibits strong generalization capability on unseen animals.
- PGP-SAM: Prototype-Guided Prompt Learning for Efficient Few-Shot Medical Image Segmentation9
This work introduces PGP-SAM, a novel prototype-based few-shot tuning approach that uses limited samples to replace tedious manual prompts and proposes two main components: a plug-and-play contextual modulation module that integrates multi-scale information and a class-guided cross-attention mechanism that fuses prototypes and features for automatic prompt generation.
- 9
- 9
- TLRM: Task-level Relation Module for GNN-based Few-Shot Learning9
This work proposes a new relation measure method, namely the task-level relation module (TLRM), to explicitly model the task -level relation of one sample to all the others, and demonstrates that the proposed module is effective for GNN-based few-shot learning.
- Deep InterBoost networks for small-sample image classification9
Experimental results on four small-sample datasets, UIUC-Sports, LabelMe, 15Scenes and Caltech101, demonstrate that the proposed ensemble method outperforms existing ones and results from the Wilcoxon signed-rank tests show that the method is statistically significantly better than the methods compared.
- PolarAnything: Diffusion-based Polarimetric Image Synthesis8
This work introduces a diffusion-based generative framework with an effective representation strategy that preserves the fidelity of polarization properties, capable of synthesizing polarization images from a single RGB input with both photorealism and physical accuracy.
- 8
- Multimodal Conditional Information Bottleneck for Generalizable AI-Generated Image Detection8
A multimodal conditional bottleneck network is proposed to reduce feature redundancy while enhancing the discriminative power of features extracted by CLIP, thereby improving the model's generalization ability.
- 8
- 8
- Dual-attention Guided Dropblock Module for Weakly Supervised Object Localization8
The dual-attention guided dropblock module (DGDM) is proposed, which aims at learning the informative and complementary visual patterns for WSOL, which achieves new state-of-the-art localization performance.
- Channel DropBlock: An Improved Regularization Method for Fine-Grained Visual Classification8
A lightweight yet effective regularization method named Channel DropBlock (CDB), in combination with two alternative correlation metrics, to address the problem of sub-categories in fine-grained visual classification (FGVC) task.
- 8
- On the Comparisons of Decorrelation Approaches for Non-Gaussian Neutral Vector Variables8
This article extensively compare PNT with PCA and ICA through both theoretical analysis and experimental evaluations, and demonstrates the superiority of PNT for decorrelating the neutral vector variables.
- 8
- 8
- A Multi-level System for Sequential Update Summarization8
A new multi-level system which can broadcasts with useful, new, and timely sentence-length updates about a developing event and incorporates techniques from topic-level and sentence-level summarization is presented.
- 7
- DriveRX: A Vision-Language Reasoning Model for Cross-Task Autonomous Driving7
This work presents AutoDriveRL, a unified training framework that formulates autonomous driving as a structured reasoning process over four core tasks, and trains DriveRX, a cross-task reasoning VLM designed for multi-stage decision-making.
- 7
- 7
- Competing ratio loss for discriminative multi-class image classification7
The experimental results show the effectiveness and robustness of the loss function on different deep convolutional neural network architectures and different image classification tasks, such as fine-grained image classification, hard face age estimation and large-scale image classification.
- 7
- 7
- 6
- 6
- Learning Dynamic Prototypes for Visual Pattern Debiasing6
Experimental results show that the proposed prototype-based network to dynamically balance the learning of different subgroups for a given dataset outperforms the CNN-based and transformer-based state-of-the-art methods in terms of both effectiveness and fairness metrics.
- 6
- Knowledge Transfer Based Fine-Grained Visual Classification6
A orthogonal loss (OR-loss) is proposed to encourage the network to find diverse and meaningful regions and is demonstrated the superiority of the proposed method and obtain state-of-the-art (SOTA) performances on three popular FGVC datasets.
- IEEE Access Special Section Editorial: AI-Driven Big Data Processing: Theory, Methodology, and Applications6
This paper discusses how to analyze and process big data in a way that is more effective and reduces costs, how to discover and understand knowledge from the data, and how to generalize and transfer these discoveries into other application fields.
- 6
- 6
- Sentiment Analysis by Exploring Large Scale Web-based Chinese Short Text6
The result shows the improvement in short text sentiment analysis of the model by using the soft mapping sentiment analysis model, which has significant implications in the psychological guidance of college students.
- Mining Sequential Update Summarization with Hierarchical Text Analysis6
This paper defines the problem of sequential update summarization extraction and presents a new hierarchical update mining system which can broadcast with useful, new, and timely sentence-length updates about a developing event and proposes a novel method, which incorporates techniques from topic-level and sentence-level summarization.
- 6
- Interactive triplet attention for few-shot fine-grained image classification5
This paper shows that a simple adjustment to the existing triplet attention module (TAM) can be highly effective for few-shot fine-grained image classification, and introduces an additional interaction between the outputs of these three branches, capable of modeling the dependency across all three dimensions.
- Measurement-Constrained Sampling for Text-Prompted Blind Face Restoration5
A Measurement-Constrained Sampling (MCS) approach that enables diverse LQ face reconstructions conditioned on different textual prompts and can generate prompt-aligned results and outperforms existing BFR methods is proposed.
- 5
- Class-Aware Contrastive Learning for Fine-Grained Skeleton-Based Action Recognition5
A class-aware contrastive learning framework designed to emphasize subtle motion feature differences that enhances discriminative capability for fine-grained action recognition by refining negative sample selection in contrastive learning to prioritize samples from similar categories.
- 5
- Triple Alignment Strategies for Zero-shot Phrase Grounding under Weak Supervision5
This paper proposes a framework of zero-shot PG under weak supervision built on triple alignment strategies and shows that the proposed PG framework outperforms previous zero-shot methods and weakly-supervised methods.
- Hierarchical Visual Attribute Learning in the Wild5
This work proposes a model called the object-specific attribute relation net, which takes advantage of three types of relations among attributes - positive, negative, and hierarchical - to better facilitate attribute recognition in images and introduces several evaluation metrics for attribute hierarchy.
- Cross-layer Navigation Convolutional Neural Network for Fine-grained Visual Classification5
This paper proposes cross-layer navigation convolutional neural network for feature fusion and demonstrates the superiority of the proposed method by comparing it with other referred FGVC methods to show that this method achieves superior results.
- 5
- Channel Max Pooling for Image Classification5
A channel max pooling, which reduces the feature space by compressing multiple feature maps into one feature map via selecting the maximum values of the same locations from different feature maps, achieves good performance and significantly reduced the parameters of the neural network.
- 5
- BALSON: BAYESIAN LEAST SQUARES OPTIMIZATION WITH NONNEGATIVE L1-NORM CONSTRAINT5
With the estimated posterior distributions, the original parameters can be effectively reconstructed in polynomial fitting problems, and the BALSON framework is found to perform better than conventional methods.
- Controllable-Continuous Color Editing in Diffusion Model via Color Mapping4
A color mapping module is introduced that explicitly models the correspondence between the text embedding space and image RGB values, enabling precise color control of the generated images while maintaining semantic consistency.
- FairHuman: Boosting Hand and Face Quality in Human Image Generation with Minimum Potential Delay Fairness in Diffusion Models4
FairHuman, a multi-objective fine-tuning approach designed to enhance both global and local generation quality fairly, is proposed, which can achieve significant improvements in generating challenging local details while maintaining overall quality.
- 4
- 4
- 4
- SEA: A Combined Model for Heat Demand Prediction4
STL-ENN-ARIMA (SEA), a combined model, was proposed based on the combination of the Elman neural network and the autoregressive integrated moving average model, which are commonly applied to heat demand prediction.
- Quantitative Comparisons of Linked Color Imaging and White-Light Colonoscopy for Colorectal Polyp Analysis4
A new criterion, namely entropy of colorGradients image (ECGI), which is based on color gradients distribution and provides a comprehensive and objective evaluating indicator of the performance of colorectal images is presented.
- Image-Text Dual Model for Small-Sample Image Classification4
The proposed image-text dual model aims to utilize the text information to overcome the problem of training deep models on small-sample datasets and can achieve the highest image classification accuracy among all the referred models on LabelMe and UIUC-Sports.
- Recent advances in machine learning for non-Gaussian data processing4
The scope of this special issue of the Elsevier's Journal on Neurocomputing is to provide theoretical foundations and ground-breaking models and algorithms to solve the challenge of analyzing and understanding non-Gaussian distributed data.
- 4
- Approximating the predictive distribution of the beta distribution with the local variational method4
- Benchmarking Semantic Segmentation Models via Appearance and Geometry Attribute Editing3
An automatic data generation pipeline Gen4Seg is constructed to stress-test semantic segmentation models by generating various challenging samples with different attribute changes, and suggests the potential of generative models as effective tools for automatically analyzing segmentation models.
- 3
- Detailed Object Description With Controllable Dimensions3
A training-free object description refinement pipeline, Dimension Tailor, designed to enhance user-specified details in object descriptions, which can consistently improve the performance of the recent MLLMs.
- 3
- Caption Feature Space Regularization for Audio Captioning3
A two-stage framework for audio captioning is proposed that constructs a proxy feature space to reduce the distances between captions correlated to the same audio, and in the second stage, theproxy feature space is utilized as additional supervision to encourage the model to be optimized in the direction that benefits all the correlated captions.
- Multi-View Active Fine-Grained Recognition3
Comprehensive experiments demonstrate that the proposed method delivers a better performance-efficient trade-off than previous FGVC methods and advanced neural networks.
- Structured Dropconnect for Uncertainty Inference in Image Classification3
A structured DropConnect (SDC) framework to model the output of a deep neural network into a distribution is proposed and the entropy of the parameterized Dirichlet distribution is finally utilized for uncertainty inference.
- CMF: Cascaded Multi-Model Fusion For Referring Image Segmentation3
This work proposes a simple yet effective Cascaded Multi-modal Fusion (CMF) module, which stacks multiple atrous convolutional layers in parallel and further introduces a cascaded branch to fuse visual and linguistic features.
- Guest Editorial Special Issue on Recent Advances in Theory, Methodology, and Applications of Imbalanced Learning3
The aim of this Special Issue is to collect recent research works that focus on the theory, methodology, and applications of imbalanced learning.
- 3
- 3
- 3
- 3
- 3
- IEEE Access Special Section Editorial: Recent Advantages of Computer Vision3
With the emergence of big vision data and the development of artificial intelligence, it is important to investigate new theories, methods, and applications in computer vision.
- 3
- Generating Realistic Labelled, Weighted Random Graphs3
A Bayesian Variational Inference approach is proposed, which yields an accurate estimation while keeping computation times tractable and comparing the approach to state-of-the-art random labelled graph generators and an earlier approach based on Gaussian Mixture Models.
- 3
- BG-NMF : a variational Bayesian NMF model for bounded support data3
This dissertation studies several estimation methods for the non-Gaussian distributions and proposes a variational Bayesian Nonnegative Matrix Factorization (NMF) forbounded support data and a Variational Inference framework based method for approximating the parameter distribution.
- 3
- 2
- Multi-View Knowledge Guided Semantic Prototype Learning for Generalized Zero-Shot Action Recognition2
- Harnessing Caption Detailness for Data-Efficient Text-to-Image Generation2
A new metric to estimate caption detailness based on two aspects: image coverage rate (ICR), which evaluates whether the caption covers all regions/objects in the image, and average object detailness (AOD), which quantifies the detailness of each object's description is proposed.
- 2
- Hierarchical Prompting for Diffusion Classifiers2
The method first proposes a hierarchical evaluation strategy, leveraging prior class tree taxonomy to reduce unnecessary class modeling, and employs prompt learning, a parameter-efficient technique, to adapt downstream task-specific knowledge into the conditional text embedding.
- M4Fog: A Global Multi-Regional, Multi-Modal, and Multi-Stage Dataset for Marine Fog Detection and Forecasting to Bridge Ocean and Atmosphere2
This work presents the most comprehensive marine fog detection and forecasting dataset to date, named M4Fog, and defines and explored three meaningful tracks with multi-metric evaluation systems: static or dynamic marine fog detection, and spatio-temporal forecasting for cloud images.
- 2
- 2
- 2
- 2
- A Fast Scheduling Method for Massive Oblique Photography 3D Models2
Experiments show that the fast scheduling method provided is efficient and feasible and can meet the visualization and real-time interaction requirements of massive urban 3D models.
- 2
- 2
- 2
- Your "Labrador" is My "Dog": Fine-Grained, or Not.2
The key intuition that: coarse-level label prediction exacerbates fine-grained feature learning, yet fine-level feature betters the learning of coarse- level classifier is discovered, which enables the design of a very simple albeit surprisingly effective solution to a new problem.
- 2
- BARE: Towards Bias-Aware and Reasoning-Enhanced One-Tower Visual Grounding1
BARE introduces a mechanism that preserves modality-specific features and constructs referential semantics through three novel modules: 1) language salience modulator; 2) visual bias correction; and 3) referential relationship enhancement, which jointly mitigate multimodal distractions and enhance referential comprehension.
- AutoRed: A Free-form Adversarial Prompt Generation Framework for Automated Red Teaming1
This work proposes AutoRed, a free-form adversarial prompt generation framework that removes the need for seed instructions and achieves higher attack success rates and better generalization than existing baselines.
- 1
- 1
- Integrating Phenological Priors with Deep Spatio-Temporal Features for tree species mapping1
A novel model, PTSViT, is proposed, which integrates phenological information with deep spatio-temporal features to address limitations of SITS, and surpasses previous approaches across all evaluation metrics, demonstrating the value of integrating prior knowledge for automated, accurate tree species mapping.
- 1
- 1
- VoiceAgentEval: A Dual-Dimensional Benchmark for Expert-Level Intelligent Voice-Agent Evaluation of Xbench's Professional-Aligned Series1
This work proposes OutboundEval, a comprehensive benchmark for evaluating large language models (LLMs) in expert-level intelligent outbound calling scenarios, and establishes a practical, extensible, and domain-oriented standard for benchmarking LLMs in professional applications.
- 1
- 1
- PMNI: Pose-free Multi-view Normal Integration for Reflective and Textureless Surface Reconstruction1
Experimental results show that the PMNI (Pose-free Multi-view Normal Integration) method achieves state-of-the-art performance in the reconstruction of reflective surfaces, even without reliable initial camera poses.
- Polyp-E: Benchmarking the Robustness of Deep Segmentation Models via Polyp Editing1
This work performs attribute editing on real polyps and builds a new dataset named Polyp-E, which boasts exceptional realism, to the extent that clinical experts find it challenging to discern them from real data.
- BlipFill: Completing Authentic Images with BLIP-enhanced Subject Representation1
This work presents a novel authentic image completion approach, which adopts a BLIP-driven memorizer to preserve subject-related information in terms of a multimodal representation and guides the fine-tuning of the inpainting/outpainting model.
- 1
- 1
- 1
- Vision-Language Assisted Attribute Learning1
This paper uses the available vision-language knowledge to explicitly disclose the missing labels for enhancing model learning, and predicts the likelihood of each missing attribute label assisted by an off-the-shelf vision-language model and randomly selects to ignore those with high scores in training.
- SLDT: Sequential Latent Document Transformer for Multilingual Document-based Dialogue1
This work proposes the application of the SLDT method, which treats passage-knowledge selection as a sequential decision process rather than a single-step decision process, and achieves winner 3rd in dialdoc 2023 and validated the effectiveness of the method on other datasets.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- Guest Editorial: Special issue on deep learning with small samples1
This special issue collects the latest developments in various aspects of deep learning with small samples and provides a concise review on recent representative methods in few-shot meta-learning, which are categorised into four branches according to their technical characteristics.
- Channel Attention with Embedding Gaussian Process: A Probabilistic Methodology.1
A Gaussian process embedded channel attention (GPCA) module is proposed and interpret the channel attention intuitively and reasonably in a probabilistic way and can improve the accuracies of image classification on four widely used datasets.
- Object-Oriented Video Captioning with Temporal Graph and Prior Knowledge Building.1
The video-based object-oriented video captioning (OVC)-Net is designed to reliably analyze the activities along time with only visual features and capture the vision-language connections under small datasets stably and exhibits the ability of precisely describing the concurrent objects and their activities in details.
- 1
- Global Context Enhanced Multi-modal Fusion for Referring Image Segmentation1
A global fusion network (GFNet), which is composed of visual guided global fusion module and language guidedglobal fusion module that outperforms state-of-the-art methods in referring image segmentation and introduces a channel-wise self-gate on visual-language concatenated features.
- 1
- 1
- 1
- 1
- Activation Force-based Air Pollution Observation Station Clustering1
This paper model the connections of the pollution sources by AF so that the relationship among the observation stations and the coincidence of the transmission links can be modeled effectively and conducts clustering of the air pollution sources via modularity measurement.
- –
- CoDA: Co-adaptive Dual-path Alignment for Vision-Language Models–
In CoDA, a new adaptation framework that explicitly disentangles and coordinates cross-modal semantic alignment and intra-modal structural consistency is proposed, and it is shown that CoDA outperforms state-of-the-art parameter-efficient methods, particularly under few-shot learning and distribution-shift scenarios.
- –
- –
- –
- FlashNormal: Detailed Surface Normal Estimation from Flash and No-Flash Images–
This work proposes FlashNormal, a diffusion-based surface normal estimator from flash/no-flash image pairs that takes advantage of flash-induced shading variations, and leverages curvature-guided detail enhancement strategy, improving surface detail recovery and mitigating shape-reflectance ambiguity effectively.
- –
- Attention-Enhanced Cross-Modality Alignment for Adapting Vision-Language Models–
An attention-enhanced cross-modality alignment network, which includes an adaptive channel attention (ACA) module and a cross-modal measurement (CMM) module, which outperforms state-of-the-art methods on two representative tasks: base-to-novel generalization and cross-dataset evaluation.
- –
- –
- –
- –
- –
- –
- –
- From Simple to Professional: A Combinatorial Controllable Image Captioning Agent–
The Controllable Image Captioning Agent is an innovative system designed to bridge the gap between user simplicity and professional-level outputs in image captioning tasks, leveraging multimodal large language models (MLLMs) and external tools such as object detection tool and search engines.
- –
- –
- –
- Linear response of entanglement entropy to $T\bar{T}$ in massive QFTs–
It is shown that both for massive scalar and Dirac fermion, the leading order correction to the entanglement entropy of half space comes from the boundary of the entangling surface.
- –
- –
- –
- –
- –
- Generalisations of stochastic supervision models–
This paper provides four generalisations of stochastic supervision models, extending them to asymmetric assessments, multiple classes, feature-dependent assessments and multi-modal classes, respectively, and derives four new EM algorithms.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
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 2026-10-14. 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.