arXiv subjects

Bowen Tang

Publications and source records attributed to Bowen Tang.

12 recordsLinked to original sources

arXiv2026

GOTS: Greedy Orthogonal Token Selection for High-Resolution Vision-Language Models

Modern vision-language models (VLMs) increasingly rely on dynamic or high-resolution visual encoding, producing thousands of visual tokens that substantially increase downstream language-model inference cost. Existing token-reduction methods assess token utility through token-wise importance, query relevance, coverage, pairwise diversity, or subset-level objectives. Our key insight is to view visual token reduction through selected-span complementarity: instead of scoring a token in isolation or through pairwise relations, we assess how much of its feature is orthogonal to the span of the already retained subset. Based on this perspective, we propose Greedy Orthogonal Token Selection (GOTS), a training-free and query-agnostic method. At each step, GOTS selects the token with the largest residual energy orthogonal to the current retained span. This rule exactly maximizes the one-step augmented Gram determinant among candidate additions, giving each greedy step a precise local geometric guarantee for subset expansion. Across five high-resolution VLM backbones from the Qwen-VL and InternVL families and eleven diverse benchmarks, GOTS achieves higher average performance retention than the strongest evaluated baselines, and a controlled OCRBench study shows that it reduces model-side time-to-first-token after accounting for selection overhead. Code is available at https://github.com/newLLing/GOTS.

cs.AI↗

arXiv2025

Q-Ponder: A Unified Training Pipeline for Reasoning-based Visual Quality Assessment

Recent studies demonstrate that multimodal large language models (MLLMs) can proficiently evaluate visual quality through interpretable assessments. However, existing approaches typically treat quality scoring and reasoning descriptions as separate tasks with disjoint optimization objectives, leading to a trade-off: models adept at quality reasoning descriptions struggle with precise score regression, while score-focused models lack interpretability. This limitation hinders the full potential of MLLMs in visual quality assessment, where accuracy and interpretability should be mutually reinforcing. To address this, we propose a unified two-stage training framework comprising a cold-start stage and a reinforcement learning-based fine-tuning stage. Specifically, in the first stage, we distill high-quality data from a teacher model through expert-designed prompts, initializing reasoning capabilities via cross-entropy loss supervision. In the second stage, we introduce a novel reward with Group Relative Policy Optimization (GRPO) to jointly optimize scoring accuracy and reasoning consistency. We designate the models derived from these two stages as Q-Ponder-CI and Q-Ponder. Extensive experiments show that Q-Ponder achieves state-of-the-art (SOTA) performance on quality score regression benchmarks, delivering up to 6.5% higher SRCC on cross-domain datasets. Furthermore, Q-Ponder significantly outperforms description-based SOTA models, including its teacher model Qwen-2.5-VL-72B, particularly in description accuracy and reasonableness, demonstrating the generalization potential over diverse tasks.

cs.CV↗

arXiv2025

ALOHA: Empowering Multilingual Agent for University Orientation with Hierarchical Retrieval

The rise of Large Language Models~(LLMs) revolutionizes information retrieval, allowing users to obtain required answers through complex instructions within conversations. However, publicly available services remain inadequate in addressing the needs of faculty and students to search campus-specific information. It is primarily due to the LLM's lack of domain-specific knowledge and the limitation of search engines in supporting multilingual and timely scenarios. To tackle these challenges, we introduce ALOHA, a multilingual agent enhanced by hierarchical retrieval for university orientation. We also integrate external APIs into the front-end interface to provide interactive service. The human evaluation and case study show our proposed system has strong capabilities to yield correct, timely, and user-friendly responses to the queries in multiple languages, surpassing commercial chatbots and search engines. The system has been deployed and has provided service for more than 12,000 people.

cs.CL↗

arXiv2024

D$^4$M: Dataset Distillation via Disentangled Diffusion Model

Dataset distillation offers a lightweight synthetic dataset for fast network training with promising test accuracy. To imitate the performance of the original dataset, most approaches employ bi-level optimization and the distillation space relies on the matching architecture. Nevertheless, these approaches either suffer significant computational costs on large-scale datasets or experience performance decline on cross-architectures. We advocate for designing an economical dataset distillation framework that is independent of the matching architectures. With empirical observations, we argue that constraining the consistency of the real and synthetic image spaces will enhance the cross-architecture generalization. Motivated by this, we introduce Dataset Distillation via Disentangled Diffusion Model (D$^4$M), an efficient framework for dataset distillation. Compared to architecture-dependent methods, D$^4$M employs latent diffusion model to guarantee consistency and incorporates label information into category prototypes. The distilled datasets are versatile, eliminating the need for repeated generation of distinct datasets for various architectures. Through comprehensive experiments, D$^4$M demonstrates superior performance and robust generalization, surpassing the SOTA methods across most aspects.

cs.CV↗

arXiv2024

Crimson: Empowering Strategic Reasoning in Cybersecurity through Large Language Models

We introduces Crimson, a system that enhances the strategic reasoning capabilities of Large Language Models (LLMs) within the realm of cybersecurity. By correlating CVEs with MITRE ATT&CK techniques, Crimson advances threat anticipation and strategic defense efforts. Our approach includes defining and evaluating cybersecurity strategic tasks, alongside implementing a comprehensive human-in-the-loop data-synthetic workflow to develop the CVE-to-ATT&CK Mapping (CVEM) dataset. We further enhance LLMs' reasoning abilities through a novel Retrieval-Aware Training (RAT) process and its refined iteration, RAT-R. Our findings demonstrate that an LLM fine-tuned with our techniques, possessing 7 billion parameters, approaches the performance level of GPT-4, showing markedly lower rates of hallucination and errors, and surpassing other models in strategic reasoning tasks. Moreover, domain-specific fine-tuning of embedding models significantly improves performance within cybersecurity contexts, underscoring the efficacy of our methodology. By leveraging Crimson to convert raw vulnerability data into structured and actionable insights, we bolster proactive cybersecurity defenses.

cs.CR↗

arXiv2024

Data-Free Generalized Zero-Shot Learning

Deep learning models have the ability to extract rich knowledge from large-scale datasets. However, the sharing of data has become increasingly challenging due to concerns regarding data copyright and privacy. Consequently, this hampers the effective transfer of knowledge from existing data to novel downstream tasks and concepts. Zero-shot learning (ZSL) approaches aim to recognize new classes by transferring semantic knowledge learned from base classes. However, traditional generative ZSL methods often require access to real images from base classes and rely on manually annotated attributes, which presents challenges in terms of data restrictions and model scalability. To this end, this paper tackles a challenging and practical problem dubbed as data-free zero-shot learning (DFZSL), where only the CLIP-based base classes data pre-trained classifier is available for zero-shot classification. Specifically, we propose a generic framework for DFZSL, which consists of three main components. Firstly, to recover the virtual features of the base data, we model the CLIP features of base class images as samples from a von Mises-Fisher (vMF) distribution based on the pre-trained classifier. Secondly, we leverage the text features of CLIP as low-cost semantic information and propose a feature-language prompt tuning (FLPT) method to further align the virtual image features and textual features. Thirdly, we train a conditional generative model using the well-aligned virtual image features and corresponding semantic text features, enabling the generation of new classes features and achieve better zero-shot generalization. Our framework has been evaluated on five commonly used benchmarks for generalized ZSL, as well as 11 benchmarks for the base-to-new ZSL. The results demonstrate the superiority and effectiveness of our approach. Our code is available in https://github.com/ylong4/DFZSL

cs.CV↗

arXiv2023

Emergence of superconductivity near 11 K by suppressing the 3-fold helical-chain structure in noncentrosymmetric HgS

The trigonal ${\alpha}$-HgS has a 3-fold helical chain structure, and is in form of a noncentrosymmetric $P3_121$ phase, known as the cinnabar phase. However, under pressure, the helical chains gradually approach and connect with each other, finally reconstructing into a centrosymmetric NaCl structure at 21 GPa. Superconductivity emerges just after this helical-nonhelical structural transition. The maximum critical temperature ($T_c$) reaches 11 K at 25.4 GPa, $T_c$ decreases with further compression, and is still 3.5 K at 44.8 GPa. Furthermore, the $T_c$-critical magnetic field ($B_{c2}$) relation exhibits multi-band features, with a $B_{c2}$ of 5.65 T at 0 K by two-band fitting. Raman spectra analysis demonstrates that phonon softening plays a key role in structural transition and the emergence of superconductivity. It is noted that HgS is the first reported IIB group metal sulfide superconductor and the only NaCl-type metal sulfide superconductor with a $T_c$ above 10 K. This work will inspire the exploration of superconductivity in other chiral systems and will extend our understanding of the versatile behavior in such kinds of materials.

cond-mat.supr-con↗

arXiv2023

SPECWANDS: An Efficient Priority-based Scheduler Against Speculation Contention Attacks

Transient Execution Attacks (TEAs) have gradually become a major security threat to modern high-performance processors. They exploit the vulnerability of speculative execution to illegally access private data, and transmit them through timing-based covert channels. While new vulnerabilities are discovered continuously, the covert channels can be categorised to two types: 1) Persistent Type, in which covert channels are based on the layout changes of buffering, e.g. through caches or TLBs; 2) Volatile Type, in which covert channels are based on the contention of sharing resources, e.g. through execution units or issuing ports. The defenses against the persistent-type covert channels have been well addressed, while those for the volatile-type are still rather inadequate. Existing mitigation schemes for the volatile type such as Speculative Compression and Time-Division-Multiplexing will introduce significant overhead due to the need to stall the pipeline or to disallow resource sharing. In this paper, we look into such attacks and defenses with a new perspective, and propose a scheduling-based mitigation scheme, called SPECWANDS. It consists of three priority-based scheduling policies to prevent an attacker from transmitting the secret in different contention situations. SPECWANDS not only can defend against both inter-thread and intra-thread based attacks, but also can keep most of the performance benefit from speculative execution and resource-sharing. We evaluate its runtime overhead on SPEC 2017 benchmarks and realistic programs. The experimental results show that SPECWANDS has a significant performance advantage over the other two representative schemes.

cs.CR↗

arXiv2022

FuncFooler: A Practical Black-box Attack Against Learning-based Binary Code Similarity Detection Methods

The binary code similarity detection (BCSD) method measures the similarity of two binary executable codes. Recently, the learning-based BCSD methods have achieved great success, outperforming traditional BCSD in detection accuracy and efficiency. However, the existing studies are rather sparse on the adversarial vulnerability of the learning-based BCSD methods, which cause hazards in security-related applications. To evaluate the adversarial robustness, this paper designs an efficient and black-box adversarial code generation algorithm, namely, FuncFooler. FuncFooler constrains the adversarial codes 1) to keep unchanged the program's control flow graph (CFG), and 2) to preserve the same semantic meaning. Specifically, FuncFooler consecutively 1) determines vulnerable candidates in the malicious code, 2) chooses and inserts the adversarial instructions from the benign code, and 3) corrects the semantic side effect of the adversarial code to meet the constraints. Empirically, our FuncFooler can successfully attack the three learning-based BCSD models, including SAFE, Asm2Vec, and jTrans, which calls into question whether the learning-based BCSD is desirable.

cs.CR↗

arXiv2022

Pre-training of Equivariant Graph Matching Networks with Conformation Flexibility for Drug Binding

The latest biological findings observe that the traditional motionless 'lock-and-key' theory is not generally applicable because the receptor and ligand are constantly moving. Nonetheless, remarkable changes in associated atomic sites and binding pose can provide vital information in understanding the process of drug binding. Based on this mechanism, molecular dynamics (MD) simulations were invented as a useful tool for investigating the dynamic properties of a molecular system. However, the computational expenditure limits the growth and application of protein trajectory-related studies, thus hindering the possibility of supervised learning. To tackle this obstacle, we present a novel spatial-temporal pre-training method based on the modified Equivariant Graph Matching Networks (EGMN), dubbed ProtMD, which has two specially designed self-supervised learning tasks: an atom-level prompt-based denoising generative task and a conformation-level snapshot ordering task to seize the flexibility information inside MD trajectories with very fine temporal resolutions. The ProtMD can grant the encoder network the capacity to capture the time-dependent geometric mobility of conformations along MD trajectories. Two downstream tasks are chosen, i.e., the binding affinity prediction and the ligand efficacy prediction, to verify the effectiveness of ProtMD through linear detection and task-specific fine-tuning. We observe a huge improvement from current state-of-the-art methods, with a decrease of 4.3% in RMSE for the binding affinity problem and an average increase of 13.8% in AUROC and AUPRC for the ligand efficacy problem. The results demonstrate valuable insight into a strong correlation between the magnitude of conformation's motion in the 3D space (i.e., flexibility) and the strength with which the ligand binds with its receptor.

cs.CE↗

arXiv2022

LighTN: Light-weight Transformer Network for Performance-overhead Tradeoff in Point Cloud Downsampling

Compared with traditional task-irrelevant downsampling methods, task-oriented neural networks have shown improved performance in point cloud downsampling range. Recently, Transformer family of networks has shown a more powerful learning capacity in visual tasks. However, Transformer-based architectures potentially consume too many resources which are usually worthless for low overhead task networks in downsampling range. This paper proposes a novel light-weight Transformer network (LighTN) for task-oriented point cloud downsampling, as an end-to-end and plug-and-play solution. In LighTN, a single-head self-correlation module is presented to extract refined global contextual features, where three projection matrices are simultaneously eliminated to save resource overhead, and the output of symmetric matrix satisfies the permutation invariant. Then, we design a novel downsampling loss function to guide LighTN focuses on critical point cloud regions with more uniform distribution and prominent points coverage. Furthermore, We introduce a feed-forward network scaling mechanism to enhance the learnable capacity of LighTN according to the expand-reduce strategy. The result of extensive experiments on classification and registration tasks demonstrates LighTN can achieve state-of-the-art performance with limited resource overhead.

cs.CV↗

arXiv2021

SpecBox: A Label-Based Transparent Speculation Scheme Against Transient Execution Attacks

Speculative execution techniques have been a cornerstone of modern processors to improve instruction-level parallelism. However, recent studies showed that this kind of techniques could be exploited by attackers to leak secret data via transient execution attacks, such as Spectre. Many defenses are proposed to address this problem, but they all face various challenges: (1) Tracking data flow in the instruction pipeline could comprehensively address this problem, but it could cause pipeline stalls and incur high performance overhead; (2) Making side effect of speculative execution imperceptible to attackers, but it often needs additional storage components and complicated data movement operations. In this paper, we propose a label-based transparent speculation scheme called SpecBox. It dynamically partitions the cache system to isolate speculative data and non-speculative data, which can prevent transient execution from being observed by subsequent execution. Moreover, it uses thread ownership semaphores to prevent speculative data from being accessed across cores. In addition, SpecBox also enhances the auxiliary components in the cache system against transient execution attacks, such as hardware prefetcher. Our security analysis shows that SpecBox is secure and the performance evaluation shows that the performance overhead on SPEC CPU 2006 and PARSEC-3.0 benchmarks is small.

cs.CR↗