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
- Vijay ChandrasekharSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works36 from public data
- 230
- CRAFT: Cross-Attentional Flow Transformer for Robust Optical Flow149
This work proposes a new architecture “CRoss-Attentional Flow Trans-former” (CRAFT), aiming to revitalize the correlation volume computation, and designed an image shifting attack that shifts input images to generate large artificial motions.
- Hardware-Aware Softmax Approximation for Deep Neural Networks34
This work focuses on optimizing the inference of non-linear functions in DNNs, with hardware constraints, while recent efforts mainly focused on linear functions in convolutional (Conv) or fully connected (FC) layers.
- One of a Kind31
This work is the first that explores how social curation can help in content-based social media technologies, taking user profiling as an example and proposes a new deep learning strategy called multi-task convolutional neural network (mtCNN) to learn profile models and profile-related visual features simultaneously.
- Efficient Joint Optimization of Layer-Adaptive Weight Pruning in Deep Neural Networks26
A novel layer-adaptive weight-pruning approach for Deep Neural Networks (DNNs) that addresses the challenge of optimizing the output distortion minimization while adhering to a target pruning ratio constraint is proposed.
- From Algorithm to Hardware: A Survey on Efficient and Safe Deployment of Deep Neural Networks25
A comprehensive survey on recent research toward the goal of high-performance, cost-efficient, and safe deployment of DNNs to provide a big picture of efficient DNNs from algorithm to hardware accelerators and security perspectives.
- Exploiting Temporal State Space Sharing for Video Semantic Segmentation17
A Temporal Video State Space Sharing (TV3S) architecture to leverage Mamba state space models for temporal feature sharing, which features a selective gating mechanism that efficiently propagates relevant information across video frames, eliminating the need for a memory-heavy feature pool.
- 17
- 16
- 13
- 8
- 7
- RDO-Q: Extremely Fine-Grained Channel-Wise Quantization via Rate-Distortion Optimization7
This work forms the quantization of deep neural networks as a rate-distortion optimization problem, and presents an ultra-fast algorithm to search the bit allocation of channels, which has only linear time complexity and can find the optimal bit allocation within a few minutes on CPU.
- Evaluating SAM2 for Video Semantic Segmentation5
This paper explores the extension of SAM2 for VSS, focusing on two primary approaches and highlighting firsthand observations and common challenges faced during this process, and suggests that leveraging SAM2 enhances overall performance in VSS, primarily due to its precise predictions of object boundaries.
- 5
- 5
- 5
- 4
- DM3D: Distortion-Minimized Weight Pruning for Lossless 3D Object Detection4
This paper proposes a novel post-training weight pruning scheme for 3D object detection that is orthogonal to all existing point cloud sparsifying methods, and introduces a lightweight scheme to efficiently acquire Hessian information, and subsequently perform dynamic programming to solve the layer-wise sparsity.
- 4
- 3
- 2
- 2
- 2
- 1
- Compress Then Adapt? No, Do It Together via Task-aware Union of Subspaces–
JACTUS explicitly mitigates the potential misalignment between the compressed subspace and downstream objectives by coupling the directions preserved for compression with those required for adaptation, yielding a deployable low-rank model that avoids retaining full frozen weights while enabling fast and robust tuning.
- –
- –
- –
- –
- –
- –
- Joint Architecture-Token-Bitwidth Multi-Axis Optimization of Vision Transformers for Semiconductor IC Packaging–
This paper presents one of the first holistic frameworks that jointly optimizes three complementary axes: architecture, token, and bit-width in ViTs and is among the earliest works to jointly optimize architecture, token, and bit-width dimensions in ViTs.
- Characterizing Detectability in 3DGS Poisoning: A Stage-wise Benchmark–
Poison-3DGS is introduced, a benchmark for stage-wise characterization of poisoning detection in 3DGS and the first systematic characterization of stage-dependent detectability in 3DGS, offering a foundation for future research on robust and reliable 3DGS systems.
- –
- Event-Based Probabilistic Embedding for POI Recommendation–
A probabilistic embedding model called Topic And Region Embedding (TARE), which embeds events by simulating the users’ decision-making process and achieves better performance in recommendation accuracy than existing state-of-the-art methods.
Publication data from OpenAlex; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-11. 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.