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
- Steven C. H. HoiSuggested from co-authorship
- Mohan KankanhalliSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works52 from public data
- 468
- 417
- CoMatch: Semi-supervised Learning with Contrastive Graph Regularization374
CoMatch is a new semi-supervised learning method that unifies dominant approaches and addresses their limitations, and achieves substantial accuracy improvements on the label-scarce CIFAR-10 and STL-10.
- The Devil Is in Classification: A Simple Framework for Long-Tail Instance Segmentation219
This work systematically investigates performance drop of the state-of-the-art two-stage instance segmentation model Mask R-CNN on the recent long-tail LVIS dataset, and unveils that a major cause is the inaccurate classification of object proposals.
- A Minimalist Approach to LLM Reasoning: from Rejection Sampling to Reinforce146
This work revisits GRPO from a reinforce-like algorithm perspective and proposes Reinforce-Rej, a minimal extension of policy gradient that filters both entirely incorrect and entirely correct samples, serving as a lightweight yet effective alternative to more complex RL algorithms.
- 127
- Reward-Guided Speculative Decoding for Efficient LLM Reasoning120
Reward-Guided Speculative Decoding is introduced, a novel framework aimed at improving the efficiency of inference in large language models (LLMs) by incorporating a controlled bias to prioritize high-reward outputs in contrast to existing speculative decoding methods that enforce strict unbiasedness.
- MCP-Universe: Benchmarking Large Language Models with Real-World Model Context Protocol Servers107
MCP-Universe is introduced, the first comprehensive benchmark specifically designed to evaluate LLMs in realistic and hard tasks through interaction with real-world MCP servers, and implements execution-based evaluators, including format evaluators for agent format compliance, static evaluators for time-invariant content matching, and dynamic evaluators that automatically retrieve real-time ground truth for temporally sensitive tasks.
- 90
- 79
- 52
- Attention Transfer from Web Images for Video Recognition48
This work proposes a novel approach to transfer knowledge from image domain to video domain, and designs a novel Siamese EnergyNet structure to learn energy functions on the attention maps by jointly optimizing two loss functions, such that the attention map corresponding to a ground truth concept would have higher energy.
- Deep Reinforcement Learning in Soft Viscoelastic Actuator of Dielectric Elastomer44
This letter proposes a model-free method for control of DEAs, based on deep reinforcement learning that is robust to achieve accurate control for the DEAs of different configurations, different prestretches, and at different times.
- Scalable Chain of Thoughts via Elastic Reasoning42
This work proposes Elastic Reasoning, a novel framework for scalable chain of thoughts that explicitly separates reasoning into two phases--thinking and solution--with independently allocated budgets, and demonstrates that it performs robustly under strict budget constraints, while incurring significantly lower training cost than baseline methods.
- Removal behavior and key drivers of antibiotic resistance genes in two full-scale leachate treatment plants42
This study provides new insights into the mechanisms of ARGs removal and dissemination behavior in industrial-scale LTPs and indicated that bacterial composition, MGEs and basic water properties were the key drivers shaping ARGs dynamics in the raw leachate, biological system and filtration system, respectively.
- Visual Social Relationship Recognition32
A dual-glance model for social relationship recognition, where the first glance fixates at the person of interest and the second glance deploys attention mechanism to exploit contextual cues, is proposed.
- 26
- Empowering Time Series Analysis with Synthetic Data: A Survey and Outlook in the Era of Foundation Models21
This survey provides a comprehensive review of synthetic data for TSFMs and TSLLMs, analyzing data generation strategies, their role in model pretraining, fine-tuning, and evaluation, and identifying future research directions.
- Improving Tail-Class Representation with Centroid Contrastive Learning20
Interpolative centroid contrastive learning (ICCL) is proposed to improve long-tailed representation learning and shows a significant accuracy gain of 2.8% on the iNaturalist 2018 dataset with a real-world long-tails distribution.
- Weakly-Supervised Multi-Person Action Recognition in 360° Videos20
This work proposes a weakly-supervised method based on multiinstance multi-label learning, which trains the model to recognize and localize multiple actions in a video using only video-level action labels as supervision.
- VideoAutoArena: An Automated Arena for Evaluating Large Multimodal Models in Video Analysis through User Simulation19
VideoAutoArena is introduced, an arena-style benchmark inspired by LMSYS Chatbot Arena’s framework, designed to automatically assess LMMs’ video analysis abilities, and introduces a fault-driven evolution strategy, progressively increasing question complexity to push models toward handling more challenging video analysis scenarios.
- Masked Unsupervised Self-training for Label-free Image Classification18
This paper proposes Masked Unsupervised Self-Training (MUST), a new approach which leverages two different and complimentary sources of supervision: pseudo-labels and raw images, and achieves a zero-shot top-1 accuracy of 77.7% on ImageNet using ViT-B, +9.4% higher than CLIP.
- A review of fiber-optic corrosion sensor in civil engineering18
The mechanism of metal corrosion is introduced, and several common methods for detecting optical fiber corrosion sensors are presented, and the latest progress of optical Fiber corrosion sensors in recent years is described.
- ARVMEC: Adaptive Recommendation of Virtual Machines for IoT in Edge–Cloud Environment16
This article proposes ARVMEC, Adaptive Recommendation of Virtual Machines for IoT in Edge-Cloud Environment, which can always provide users with the best VM recommendation according to their own budget or deadline constraints.
- 15
- What Are We Measuring When We Evaluate Large Vision-Language Models? An Analysis of Latent Factors and Biases14
A large-scale transfer learning experiment aimed at discovering latent VL skills from data reveals interesting characteristics that have important implications for test suite design and contributes to the design of balanced and broad-coverage vision-language evaluation methods.
- 14
- SERAC3: Smart and economical resource allocation for big data clusters in community clouds14
SERAC3 is a system that can automatically extract representative workloads from incoming big data analysis jobs, smartly decide an optimal configuration for each job, and adjust its assigning strategy in a quasi-realtime mode to allocate resources smartly and economically for big data clusters in community clouds.
- SHE: A Generic Framework for Data Stream Mining over Sliding Windows13
This paper proposes the Sliding Hardware Estimator (SHE), which is a generic framework that extends existing fixed window algorithms to sliding windows on hardware platforms, and shows that, when compared with the state-of-the-art, SHE reduces the error by up to 100 times in membership queries.
- PipeCache: High Hit Rate Rule-Caching Scheme Based on Multi-Stage Cache Tables11
A new design named PipeCache is proposed that solves the problem of MFTs rule-caching scheme based on wildcard-match by logically split the TCAM resources, and assign them to each flow table according to the size of each flowtable.
- 9
- MsBV: A Memory Compression Scheme for Bit-Vector-Based Classification Lookup Tables9
The current work proposes the FPGA-based Bit-Vector algorithms in packet classification, which has the advantages of determining latency and high throughput, and a bit-vector-based memory compression scheme named Memory-shared Bit- Vector (MsBV), which can reduce memory consumption and ensure the correctness of packet classification.
- GradMix: Multi-source Transfer across Domains and Tasks9
This work proposes GradMix, a model-agnostic method applicable to any model trained with gradient-based learning rule, to transfer knowledge via gradient descent by weighting and mixing the gradients from all sources during training, and conducts MS-DTT experiments on two tasks: digit recognition and action recognition.
- Efficient Text-to-Code Retrieval with Cascaded Fast and Slow Transformer Models8
An efficient and effective text-to-code search framework with cascaded fast and slow models, in which a fast transformer encoder model is learned to optimize a scalable index for fast retrieval followed by learning a slow classification-based re-ranking model to improve the accuracy of the top K results from the fast retrieval.
- 8
- Fractured Chain-of-Thought Reasoning7
Fractured Sampling is introduced, a unified inference-time strategy that interpolates between full CoT and solution-only sampling along three orthogonal axes: the number of reasoning trajectories, the number of final solutions per trajectory, and the depth at which reasoning traces are truncated.
- 7
- 6
- RA-QoS: a robust autoencoder-based QoS predictor for highly accurate web service QoS prediction5
This work proposes a novel robust autoencoder-based QoS predictor (RA-QoS) that leverages a hybrid loss function combining bias, training bias, L1-norm and L2-norm to build a robust Autoencoder.
- 4
- Encoding the regional features for person re-identification using locality-constrained linear coding4
- An Efficient Method for Training Deep Learning Networks Distributed2
This paper proposes a hierarchical synchronous Stochastic Gradient Descent (SGD) strategy, which can make full use of hardware resources and greatly increase computational ef-ficiency and integrates the LARS algorithm into the system.
- 1
- –
- Adaptive Modality Routing for Temporal Knowledge Graph Forecasting with Heterogeneous Mixture-of-Experts–
Experiments indicate that the proposed H-MoE framework can improve forecasting performance under the evaluation setting and support heterogeneous expert specialization as a useful design direction, while the scope of the conclusions remains limited to the tested ICEWS benchmarks and implementation.
- –
- –
- –
- –
- On construction of a cloud based storage platform for teaching resource of obstetrics and gynecology–
A cloud based storage platform is built to manage the teaching resource of Obstetrics, which enables the efficiently sharing and integration of obstetrics related courses and has demonstrated to improve the work efficacy of teachers.
- Design of SAS real-time image display based on FPGA and DSP–
In order to improve the performance of real-time, stability and high-resolution display of sonar signal processing system, the article designs and realizes a kind of mobile phone screen real- time display system based on FPGA + DSP, with the SSD2828 device as the hardware core.
- –
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.