Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks86 from public data
- Learning Efficient Convolutional Networks through Network Slimming2,909
The approach is called network slimming, which takes wide and large networks as input models, but during training insignificant channels are automatically identified and pruned afterwards, yielding thin and compact models with comparable accuracy.
- DSOD: Learning Deeply Supervised Object Detectors from Scratch646
Deeply Supervised Object Detector (DSOD), a framework that can learn object detectors from scratch following the single-shot detection (SSD) framework, and one of the key findings is that deep supervision, enabled by dense layer-wise connections, plays a critical role in learning a good detector.
- ReActNet: Towards Precise Binary Neural Network with Generalized Activation Functions452
This paper proposes to generalize the traditional Sign and PReLU functions to enable explicit learning of the distribution reshape and shift at near-zero extra cost and shows that the proposed ReActNet outperforms all the state-of-the-arts by a large margin.
- Semantic Relation Reasoning for Shot-Stable Few-Shot Object Detection243
This work represents each class concept by a semantic embedding learned from a large corpus of text and proposes to augment the embeddings with a dynamic relation graph, resulting in a few-shot detector that is robust and stable to the variation of shots of novel objects.
- 226
- Partial Is Better Than All: Revisiting Fine-tuning Strategy for Few-shot Learning210
This paper proposes to transfer partial knowledge by freezing or fine-tuning particular layer(s) in the base model to control the extent of preserved transferability, and introduces an evolutionary search based method that is efficient to simultaneously locate the target layers and determine their individual learning rates.
- Nonuniform-to-Uniform Quantization: Towards Accurate Quantization via Generalized Straight-Through Estimation175
This study proposes Nonuniform-to-Uniform Quantization (N2UQ), a method that can maintain the strong representation ability of nonuniform methods while being hardware-friendly and efficient as the uniform quantization for model inference.
- Soft Anchor-Point Object Detection168
This work boosts the performance of the anchor-point detector over the key-point counterparts while maintaining the speed advantage, and proposes a simple yet effective training strategy with soft-weighted anchor points and soft-selected pyramid levels to address the false attention issue within each pyramid level and the feature selection issue across all the pyramid levels.
- TS$$^{2}$$C: Tight Box Mining with Surrounding Segmentation Context for Weakly Supervised Object Detection160
This work provides a simple approach to discover tight object bounding boxes with only image-level supervision, called Tight box mining with Surrounding Segmentation Context (TS2C), developed based on two key properties of desirable bounding box: high purity and high completeness.
- MEAL: Multi-Model Ensemble via Adversarial Learning157
This paper proposes an adversarial-based learning strategy where a block-wise training loss is defined to guide and optimize the predefined student network to recover the knowledge in teacher models, and to promote the discriminator network to distinguish teacher vs. student features simultaneously.
- Principled Instructions Are All You Need for Questioning LLaMA-1/2, GPT-3.5/4148
26 guiding principles designed to streamline the process of querying and prompting large language models are introduced and hope that this work can provide a better guide for researchers working on the prompting of large language models.
- Weakly Supervised Dense Video Captioning142
This paper focuses on a novel and challenging vision task, dense video captioning, which aims to automatically describe a video clip with multiple informative and diverse caption sentences, and proposes lexical fully convolutional neural networks with weakly supervised multi-instance multi-label learning to weakly link video regions with lexical labels.
- 136
- One-for-All: Generalized LoRA for Parameter-Efficient Fine-tuning129
Generalized LoRA is presented, an advanced approach for universal parameter-efficient fine-tuning tasks, which outperforms all previous methods in natural, specialized, and structured vision benchmarks, achieving superior accuracy with fewer parameters and computations.
- LLM360: Towards Fully Transparent Open-Source LLMs124
LLM360, an initiative to fully open-source LLMs, is presented, which advocates for all training code and data, model checkpoints, and intermediate results to be made available to the community.
- Towards Instance-Level Image-To-Image Translation109
This paper presents a simple yet effective instance-aware image-to-image translation approach (INIT), which employs the fine-grained local (instance) and global styles to the target image spatially and collects a large-scale benchmark for the new instance-level translation task.
- Open-LLM-Leaderboard: From Multi-choice to Open-style Questions for LLMs Evaluation, Benchmark, and Arena103
The Open-LLM-Leaderboard is introduced to track various LLMs' performance and reflect true capability of them, such as GPT-4o/4/3.5, Claude 3, Gemini, etc.
- Object Detection from Scratch with Deep Supervision103
Deeply Supervised Object Detectors (DSOD), an object detection framework that can be trained from scratch, is proposed based on the single-shot detection framework (SSD), and achieves consistently better results than the state-of-the-art methods with much more compact models.
- Vision Transformer Slimming: Multi-Dimension Searching in Continuous Optimization Space96
This paper explores the feasibility of finding an optimal sub-model from a vision transformer and introduces a pure vision transformer slimming (ViT-Slim) framework that can search a sub-structure from the original model end-to-end across multiple dimensions with state-of-the-art performance.
- SlimPajama-DC: Understanding Data Combinations for LLM Training88
The impacts of various data combinations on the pretraining of large language models using SlimPajama aims to understand the impacts of various data combinations and analyzes and discusses how global and local deduplications affect the performance of trained models.
- MoBiNet: A Mobile Binary Network for Image Classification77
A novel neural network architecture, namely MoBi- Net - Mobile Binary Network in which skip connections are manipulated to prevent information loss and vanishing gradient, thus facilitate the training process and results in an effectively small model while keeping the accuracy comparable to existing ones.
- 77
- A Fast Knowledge Distillation Framework for Visual Recognition70
A Fast Knowledge Distillation (FKD) framework is presented that replicates the distillation training phase and generates soft labels using the multi-crop KD approach, while training faster than ReLabel since no post-processes such as RoI align and softmax operations are used.
- Solving Missing-Annotation Object Detection with Background Recalibration Loss63
A superior solution called Background Recalibration Loss (BRL) that can automatically re-calibrate the loss signals according to the pre-defined IoU threshold and input image and is built on the one-stage detector which is faster and lighter.
- Initializing Models with Larger Ones51
Weight selection is introduced, a method for initializing smaller models by selecting a subset of weights from a pretrained larger model, which enables the transfer of knowledge from pretrained weights to smaller models.
- Binarizing MobileNet via Evolution-Based Searching48
A use of evolutionary search is proposed to facilitate the construction and training scheme when binarizing MobileNet, a compact network with separable depth-wise convolution and manipulation of group convolution to design efficient 1-Bit Convolutional Neural Networks (CNNs).
- “BNN - BN = ?”: Training Binary Neural Networks without Batch Normalization46
This work demonstrates for the first time that BNs can be completely removed from BNN training and inference regimes, and by plugging in and customizing techniques including adaptive gradient clipping, scale weight standardization, and specialized bottleneck block, a BN-free BNN is capable of maintaining competitive accuracy compared to its Bn-based counterpart.
- 45
- Sliced Recursive Transformer42
The proposed model is a novel and parameter-efficient vision transformer design that is compatible with a broad range of other designs for efficient ViT architectures and establishes significant improvement on ImageNet-1K over state-of-the-art methods while containing fewer parameters.
- Learning Object Detectors from Scratch with Gated Recurrent Feature Pyramids.38
This study proposes a recurrent feature-pyramid structure to squeeze rich spatial and semantic features into a single prediction layer that further reduces the number of parameters to learn, and is the best performed model of learning object detection from scratch.
- PointCMP: Contrastive Mask Prediction for Self-supervised Learning on Point Cloud Videos37
The proposed contrastive mask prediction (PointCMP) framework employs a two-branch structure to achieve simultaneous learning of both local and global spatiotemporal information and generates hard samples to facilitate learning representations with better discrimi-nation and generalization performance.
- Biomedical image analysis competitions: The state of current participation practice36
An international survey was designed that was issued to all participants of challenges conducted in conjunction with the IEEE ISBI 2021 and MICCAI 2021 conferences, covering participants' expertise and working environments, their chosen strategies, as well as algorithm characteristics.
- Masked Spatio-Temporal Structure Prediction for Self-supervised Learning on Point Cloud Videos33
MaST-Pre is a Masked Spatio-Temporal Structure Prediction method to capture the structure of point cloud videos without human annotations and consists of two self-supervised learning tasks that are forced to model the spatial and temporal structure in point cloud videos.
- CDTD: A Large-Scale Cross-Domain Benchmark for Instance-Level Image-to-Image Translation and Domain Adaptive Object Detection28
A large-scale cross-domain benchmark CDTD is introduced and a novel instance-level image-to-image translation approach called INIT and a gradient detach method for the domain adaptive object detection to harvest and exert dataset’s function of the instance level annotations across different domains is proposed.
- Point Contrastive Prediction with Semantic Clustering for Self-Supervised Learning on Point Cloud Videos27
This paper proposes a selection strategy to retain proper negatives and make use of high-similarity samples from other instances as positive supplements for contrastive learning at the point level, and introduces a new pretext task by achieving semantic alignment of superpoints.
- 25
- 25
- S2-BNN: Bridging the Gap Between Self-Supervised Real and 1-bit Neural Networks via Guided Distribution Calibration25
This paper presents a novel guided learning paradigm from real-valued to distill binary networks on the final prediction distribution, to minimize the loss and obtain desirable accuracy on BNNs.
- 24
- 23
- Joint Multi-Dimension Pruning via Numerical Gradient Update22
Joint multi-dimension pruning discovers a better status than previous studies that focused on individual dimensions solely, as the method is optimized collaboratively across the three dimensions in a single end-to-end training and it is more efficient than the previous exhaustive methods.
- Rethinking Image Mixture for Unsupervised Visual Representation Learning21
Despite its conceptual simplicity, it is shown empirically that with the simple solution -- image mixture, the authors can learn more robust visual representations from the transformed input, and the benefits of representations learned from this space can be inherited by the linear classification and downstream tasks.
- 21
- Time Blindness: Why Video-Language Models Can't See What Humans Can?17
A benchmark where information is encoded solely in temporal sequences of noise-like frames, mirroring natural phenomena from biological signaling to covert communication, to catalyze research in temporal pattern recognition and bridge the gap between human and machine video understanding.
- 17
- VideoMolmo: Spatio-Temporal Grounding Meets Pointing15
This work introduces VideoMolmo, a large multimodal model tailored for fine-grained spatio-temporal pointing conditioned on textual descriptions, and substantially improves spatio-temporal pointing accuracy and reasoning capability.
- Self-supervised Dataset Distillation: A Good Compression Is All You Need15
The proposed SC-DD outperforms all previous state-of-the-art supervised dataset distillation methods when employing larger models, such as SRe$^2$L, MTT, TESLA, DC, CAFE, etc., by large margins under the same recovery and post-training budgets.
- Exploring Simple and Transferable Recognition-Aware Image Processing14
This work examines simple approaches to improve machine recognition of processed images: optimizing the recognition loss directly on the image processing network or through an intermediate input transformation model, and shows that the accuracy gain can transfer to a black-box cloud model.
- 12
- 12
- Crystal: Illuminating LLM Abilities on Language and Code9
This work proposes a pretraining strategy to enhance the integration of natural language and coding capabilities within a single LLM, which includes two phases of training with appropriately adjusted code/language ratios and demonstrates remarkable capabilities in both domains.
- Transferable Recognition-Aware Image Processing9
This work proposes simple approaches to improve machine interpretability of processed images: optimizing the recognition loss directly on the image processing network or through an intermediate transforming model, and brings substantial accuracy improvement on both the same recognition model and when transferring to a different one, with minimal or no loss in the imageprocessing quality.
- FerKD: Surgical Label Adaptation for Efficient Distillation7
This work introduces a stabilized SelfMix augmentation that weakens the variation of the mixed images and corresponding soft labels through mixing similar regions within the same image, and demonstrates empirically that this method can dramatically improve the convergence speed and final accuracy.
- Joint Multi-Dimension Pruning7
Joint multi-dimension pruning is presented, a new perspective of pruning a network on three crucial aspects: spatial, depth and channel simultaneously, as this method is optimized collaboratively across the three dimensions in a single end-to-end training.
- 7
- NCMS: Towards accurate anchor free object detection through ℓ2 norm calibration and multi-feature selection5
The proposed detector, called NCMS, inserts a simple norm calibration operation between the feature pyramids and detection head to alleviate and balance the norm bias caused by feature pyramid network (FPN).
- 4
- 4
- Operation-Guided Progressive Human-to-AI Text Transformation Benchmark for Multi-Granularity AI-Text Detection3
OpAI-Bench provides a controlled testbed for analyzing whether, when, and how AI-assisted writing becomes detectable under realistic progressive editing scenarios, and reveals that mixed-authorship intermediate versions are often harder to detect than both fully human and heavily AI-edited endpoints, exposing non-monotonic detection patterns missed by existing benchmarks.
- One Last Attention for Your Vision-Language Model3
RAda employs a learned mask, obtained from a lightweight attention layer attached at the end of a VLM, to dynamically calibrate the contribution of each element in the rational matrix, enabling targeted adjustments to the final cross-modal interactions without incurring costly modifications to intermediate features.
- Semantics-Aware Patch Encoding and Hierarchical Dependency Modeling for Long-Term Time Series Forecasting3
Mixture of Universals (MoU), a novel framework designed to prevent semantic loss during patch encoding and efficiently enhance long-term dynamics through a hybrid approach, is proposed.
- 3
- Mobile-MMLU: A Mobile Intelligence Language Understanding Benchmark3
The Mobile-MMLU family offers a standardized framework for developing and comparing mobile-optimized LLMs, enabling advancements in productivity and decision-making within mobile computing environments.
- RCV2023 Challenges: Benchmarking Model Training and Inference for Resource-Constrained Deep Learning3
- 3
- 2
- 2
- Iterative object and part transfer for fine-grained recognition2
This paper proposes a nonparametric data-driven method for object and part localization that achieves better results than many state-of-the-art approaches, including a few using oracle (manually annotated) bounding boxes in the test images.
- Diff-ES: Stage-wise Structural Diffusion Pruning via Evolutionary Search1
Stage-wise structural Diff-ES is introduced, a stage-wise structural diffusion model pruning framework via Evolutionary search, which optimizes the stage-wise sparsity schedule and executes it through memory-efficient weight routing without model duplication.
- 1
- 1
- 1
- –
- Learning Robust Diffusion Models from Imprecise Supervision–
DMIS is proposed, a unified framework for training robust Diffusion Models from Imprecise Supervision from Imprecise Supervision, which is the first systematic study within diffusion models.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
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 2026-10-15. 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.