arXiv subjects

Yi Jiang

Publications and source records attributed to Yi Jiang.

At least 73 records · Page 4Linked to original sources

arXiv2025

DeepDR: an integrated deep-learning model web server for drug repositioning

Background: Identifying new indications for approved drugs is a complex and time-consuming process that requires extensive knowledge of pharmacology, clinical data, and advanced computational methods. Recently, deep learning (DL) methods have shown their capability for the accurate prediction of drug repositioning. However, implementing DL-based modeling requires in-depth domain knowledge and proficient programming skills. Results: In this application, we introduce DeepDR, the first integrated platform that combines a variety of established DL-based models for disease- and target-specific drug repositioning tasks. DeepDR leverages invaluable experience to recommend candidate drugs, which covers more than 15 networks and a comprehensive knowledge graph that includes 5.9 million edges across 107 types of relationships connecting drugs, diseases, proteins/genes, pathways, and expression from six existing databases and a large scientific corpus of 24 million PubMed publications. Additionally, the recommended results include detailed descriptions of the recommended drugs and visualize key patterns with interpretability through a knowledge graph. Conclusion: DeepDR is free and open to all users without the requirement of registration. We believe it can provide an easy-to-use, systematic, highly accurate, and computationally automated platform for both experimental and computational scientists.

cs.LG↗

arXiv2025

DetailFlow: 1D Coarse-to-Fine Autoregressive Image Generation via Next-Detail Prediction

This paper presents DetailFlow, a coarse-to-fine 1D autoregressive (AR) image generation method that models images through a novel next-detail prediction strategy. By learning a resolution-aware token sequence supervised with progressively degraded images, DetailFlow enables the generation process to start from the global structure and incrementally refine details. This coarse-to-fine 1D token sequence aligns well with the autoregressive inference mechanism, providing a more natural and efficient way for the AR model to generate complex visual content. Our compact 1D AR model achieves high-quality image synthesis with significantly fewer tokens than previous approaches, i.e. VAR/VQGAN. We further propose a parallel inference mechanism with self-correction that accelerates generation speed by approximately 8x while reducing accumulation sampling error inherent in teacher-forcing supervision. On the ImageNet 256x256 benchmark, our method achieves 2.96 gFID with 128 tokens, outperforming VAR (3.3 FID) and FlexVAR (3.05 FID), which both require 680 tokens in their AR models. Moreover, due to the significantly reduced token count and parallel inference mechanism, our method runs nearly 2x faster inference speed compared to VAR and FlexVAR. Extensive experimental results demonstrate DetailFlow's superior generation quality and efficiency compared to existing state-of-the-art methods.

cs.CV↗

arXiv2025

Sharp Stability of Solitons for the Cubic-Quintic NLS on R^2

This paper concerns with the cubic-quintic nonlinear Schrödinger equation on R^2. A family of new variational problems related to the solitons are introduced and solved. Some key monotonicity and uniqueness results are obtained. Then the orbital stability of solitons at every frequency are proved in terms of the Cazenave and Lions' argument. And classification of normalized ground states is first presented. Our results settle the questions raised by Lewin and Rota Nodari as well as Carles and Sparber.

math.AP↗

arXiv2025

Uniqueness and stability of normalized ground states for Hartree equation with a harmonic potential

The dynamic properties of normalized ground states for the Hartree equation with a harmonic potential are addressed. The existence of normalized ground state for any prescribed mass is confirmed according to mass-energy constrained variational approach. The uniqueness is shown by the strictly convex properties of the energy functional. Moreover, the orbital stability of every normalized ground state is proven in terms of the Cazenave and Lions' argument.

math.AP↗

arXiv2025

UniTok: A Unified Tokenizer for Visual Generation and Understanding

Visual generative and understanding models typically rely on distinct tokenizers to process images, presenting a key challenge for unifying them within a single framework. Recent studies attempt to address this by connecting the training of VQVAE (for autoregressive generation) and CLIP (for understanding) to build a unified tokenizer. However, directly combining these training objectives has been observed to cause severe loss conflicts. In this paper, we show that reconstruction and semantic supervision do not inherently conflict. Instead, the underlying bottleneck stems from limited representational capacity of discrete token space. Building on these insights, we introduce UniTok, a unified tokenizer featuring a novel multi-codebook quantization mechanism that effectively scales up the vocabulary size and bottleneck dimension. In terms of final performance, UniTok sets a new record of 0.38 rFID and 78.6% zero-shot accuracy on ImageNet. Besides, UniTok can be seamlessly integrated into MLLMs to unlock native visual generation capability, without compromising the understanding performance. Additionally, we show that UniTok favors cfg-free generation, reducing gFID from 14.6 to 2.5 on ImageNet 256$\times$256 benchmark. GitHub: https://github.com/FoundationVision/UniTok.

cs.CV↗

arXiv2025

Pty-Chi: A PyTorch-based modern ptychographic data analysis package

Ptychography has become an indispensable tool for high-resolution, non-destructive imaging using coherent light sources. The processing of ptychographic data critically depends on robust, efficient, and flexible computational reconstruction software. We introduce Pty-Chi, an open-source ptychographic reconstruction package built on PyTorch that unifies state-of-the-art analytical algorithms with automatic differentiation methods. Pty-Chi provides a comprehensive suite of reconstruction algorithms while supporting advanced experimental parameter corrections such as orthogonal probe relaxation and multislice modeling. Leveraging PyTorch as the computational backend ensures vendor-agnostic GPU acceleration, multi-device parallelization, and seamless access to modern optimizers. An object-oriented, modular design makes Pty-Chi highly extendable, enabling researchers to prototype new imaging models, integrate machine learning approaches, or build entirely new workflows on top of its core components. We demonstrate Pty-Chi's capabilities through challenging case studies that involve limited coherence, low overlap, and unstable illumination during scanning, which highlight its accuracy, versatility, and extensibility. With community-driven development and open contribution, Pty-Chi offers a modern, maintainable platform for advancing computational ptychography and for enabling innovative imaging algorithms at synchrotron facilities and beyond.

physics.optics↗

arXiv2025

Enhancing Adversarial Transferability with Adversarial Weight Tuning

Deep neural networks (DNNs) are vulnerable to adversarial examples (AEs) that mislead the model while appearing benign to human observers. A critical concern is the transferability of AEs, which enables black-box attacks without direct access to the target model. However, many previous attacks have failed to explain the intrinsic mechanism of adversarial transferability. In this paper, we rethink the property of transferable AEs and reformulate the formulation of transferability. Building on insights from this mechanism, we analyze the generalization of AEs across models with different architectures and prove that we can find a local perturbation to mitigate the gap between surrogate and target models. We further establish the inner connections between model smoothness and flat local maxima, both of which contribute to the transferability of AEs. Further, we propose a new adversarial attack algorithm, \textbf{A}dversarial \textbf{W}eight \textbf{T}uning (AWT), which adaptively adjusts the parameters of the surrogate model using generated AEs to optimize the flat local maxima and model smoothness simultaneously, without the need for extra data. AWT is a data-free tuning method that combines gradient-based and model-based attack methods to enhance the transferability of AEs. Extensive experiments on a variety of models with different architectures on ImageNet demonstrate that AWT yields superior performance over other attacks, with an average increase of nearly 5\% and 10\% attack success rates on CNN-based and Transformer-based models, respectively, compared to state-of-the-art attacks. Code available at https://github.com/xaddwell/AWT.

cs.CR↗

arXiv2025

CoCoA: Collaborative Chain-of-Agents for Parametric-Retrieved Knowledge Synergy

Retrieval-Augmented Generation (RAG) enhances Large Language Models (LLMs), especially for knowledge-intensive tasks. Despite its advantages, current RAG methods often struggle to fully exploit knowledge during generation. In particular, the synergy between the model's internal parametric knowledge and external retrieved knowledge remains limited. Retrieved contents may sometimes mislead generation, while certain generated content can guide the model toward more accurate outputs. In this work, we propose Collaborative Chain-of-Agents, a framework designed to enhance explicitly synergy over both parametric and retrieved knowledge. Specifically, we first introduce CoCoA-zero, a multi-agent RAG framework that first performs conditional knowledge induction and then reasons answers. Building on this, we develop CoCoA, a long-chain training strategy that synthesizes extended multi-agent reasoning trajectories from CoCoA-zero to fine-tune the LLM. This strategy enhances the model's capability to explicitly integrate and jointly leverage parametric and retrieved knowledge. Experimental results demonstrate the superiority of CoCoA in open-domain QA and multi-hop QA.

cs.CL↗

arXiv2025

QAgent: A modular Search Agent with Interactive Query Understanding

Large language models (LLMs) excel at natural language tasks but are limited by their static parametric knowledge, especially in knowledge-intensive task. Retrieval-augmented generation (RAG) mitigates this by integrating external information. However, (1) traditional RAG struggles with complex query understanding, and (2) even search agents trained with reinforcement learning (RL), despite their promise, still face generalization and deployment challenges. To address these limitations, we propose QAgent, a unified agentic RAG framework that employs a search agent for adaptive retrieval. This agent optimizes its understanding of the query through interactive reasoning and retrieval. To facilitate real-world application, we focus on modular search agent for query understanding that are plug-and-play in complex systems. Secifically, the agent follows a multi-step decision process trained with RL to maximize retrieval quality and support accurate downstream answers. We further analyze the strengths and weaknesses of end-to-end RL and propose a strategy that focuses on effective retrieval, thereby enhancing generalization in LLM applications. Experiments show QAgent excels at QA and serves as a plug-and-play module for real-world deployment.

cs.AI↗

arXiv2025

Reformulation is All You Need: Addressing Malicious Text Features in DNNs

Human language encompasses a wide range of intricate and diverse implicit features, which attackers can exploit to launch adversarial or backdoor attacks, compromising DNN models for NLP tasks. Existing model-oriented defenses often require substantial computational resources as model size increases, whereas sample-oriented defenses typically focus on specific attack vectors or schemes, rendering them vulnerable to adaptive attacks. We observe that the root cause of both adversarial and backdoor attacks lies in the encoding process of DNN models, where subtle textual features, negligible for human comprehension, are erroneously assigned significant weight by less robust or trojaned models. Based on it we propose a unified and adaptive defense framework that is effective against both adversarial and backdoor attacks. Our approach leverages reformulation modules to address potential malicious features in textual inputs while preserving the original semantic integrity. Extensive experiments demonstrate that our framework outperforms existing sample-oriented defense baselines across a diverse range of malicious textual features.

cs.LG↗

arXiv2025

Waver: Wave Your Way to Lifelike Video Generation

We present Waver, a high-performance foundation model for unified image and video generation. Waver can directly generate videos with durations ranging from 5 to 10 seconds at a native resolution of 720p, which are subsequently upscaled to 1080p. The model simultaneously supports text-to-video (T2V), image-to-video (I2V), and text-to-image (T2I) generation within a single, integrated framework. We introduce a Hybrid Stream DiT architecture to enhance modality alignment and accelerate training convergence. To ensure training data quality, we establish a comprehensive data curation pipeline and manually annotate and train an MLLM-based video quality model to filter for the highest-quality samples. Furthermore, we provide detailed training and inference recipes to facilitate the generation of high-quality videos. Building on these contributions, Waver excels at capturing complex motion, achieving superior motion amplitude and temporal consistency in video synthesis. Notably, it ranks among the Top 3 on both the T2V and I2V leaderboards at Artificial Analysis (data as of 2025-07-30 10:00 GMT+8), consistently outperforming existing open-source models and matching or surpassing state-of-the-art commercial solutions. We hope this technical report will help the community more efficiently train high-quality video generation models and accelerate progress in video generation technologies. Official page: https://github.com/FoundationVision/Waver.

cs.CV↗

arXiv2025

Spatio-Temporal Pruning for Compressed Spiking Large Language Models

Large Language Models (LLMs) present significant challenges for deployment in energy-constrained environments due to their large model sizes and high inference latency. Spiking Neural Networks (SNNs), inspired by the sparse event-driven neural processing and energy-efficient information transmission in the brain, offer a promising alternative for achieving low-power computing. Integrating the event-driven efficiency of spiking neurons with the advanced capabilities of LLMs represents a promising direction for power-efficient LLMs. This work specifically delves into the design of compressed spiking LLMs. Here, we revisit spatial and temporal pruning from the perspective of SNNs and propose a novel spatio-temporal pruning framework for Spiking LLMs to optimize computational efficiency while preserving high performance. Our spatial pruning technique reduces the number of active neurons and attention heads, effectively lowering the computational complexity of the model. Meanwhile, temporal pruning minimizes inference latency by dynamically adjusting the number of timesteps required for different layers. By combining these approaches with other compression techniques, we present the first work in the domain of Spiking LLMs to jointly explore spatial pruning, temporal pruning, extreme quantization and knowledge distillation strategies. Extensive experimental evaluation of our proposed framework for SpikingBERT on the large-scale GLUE benchmark demonstrates the efficacy of our approach in terms of computational operations and inference latency. Our approach offers a compelling solution for real-time, low-power natural language processing applications, making Spiking LLMs more practical for deployment on edge devices and in power-constrained settings.

cs.NE↗

arXiv2025

Emergent Interacting Phases in the Strong Coupling Limit of Twisted M-Valley Moiré Systems: Application to SnSe${}_2$

We construct an interacting Wannier model for both AA-stacked and AB-stacked twisted SnSe2, revealing a rich landscape of correlated quantum phases. For the AA-stacked case, the system is effectively described by a three-orbital triangular lattice model, where each orbital corresponds to a valley and exhibits an approximate one-dimensional hopping structure due to a new momentum-space non-symmorphic symmetry. By exploring the interacting phase diagram using a combination of theoretical methods, including Hartree-Fock mean-field theory and exact solutions of the spin model in certain limits, we identify several exotic quantum phases. These include a dimerized phase with finite residual entropy, valence bond solids, and quantum paramagnetism. In the AB-stacked case, the system realizes an interacting kagome lattice model, where the Wannier orbitals associated with the three valleys form three sublattices. In the strong coupling regime, we use cluster mean-field methods to demonstrate the emergence of a classical spin liquid phase due to the frustrated lattice structure. The high tunability of the moiré system, which allows control over both the filling and interaction strength (via twist angle), renders twisted SnSe2 a versatile platform for realizing a wide range of exotic correlated quantum phases.

cond-mat.str-el↗

arXiv2025

TokenFlow: Unified Image Tokenizer for Multimodal Understanding and Generation

We present TokenFlow, a novel unified image tokenizer that bridges the long-standing gap between multimodal understanding and generation. Prior research attempt to employ a single reconstruction-targeted Vector Quantization (VQ) encoder for unifying these two tasks. We observe that understanding and generation require fundamentally different granularities of visual information. This leads to a critical trade-off, particularly compromising performance in multimodal understanding tasks. TokenFlow addresses this challenge through an innovative dual-codebook architecture that decouples semantic and pixel-level feature learning while maintaining their alignment via a shared mapping mechanism. This design enables direct access to both high-level semantic representations crucial for understanding tasks and fine-grained visual features essential for generation through shared indices. Our extensive experiments demonstrate TokenFlow's superiority across multiple dimensions. Leveraging TokenFlow, we demonstrate for the first time that discrete visual input can surpass LLaVA-1.5 13B in understanding performance, achieving a 7.2\% average improvement. For image reconstruction, we achieve a strong FID score of 0.63 at 384*384 resolution. Moreover, TokenFlow establishes state-of-the-art performance in autoregressive image generation with a GenEval score of 0.55 at 256*256 resolution, achieving comparable results to SDXL.

cs.CV↗

arXiv2025

Advancing the Foundation Model for Music Understanding

The field of Music Information Retrieval (MIR) is fragmented, with specialized models excelling at isolated tasks. In this work, we challenge this paradigm by introducing a unified foundation model named MuFun for holistic music understanding. Our model features a novel architecture that jointly processes instrumental and lyrical content, and is trained on a large-scale dataset covering diverse tasks such as genre classification, music tagging, and question answering. To facilitate robust evaluation, we also propose a new benchmark for multi-faceted music understanding called MuCUE (Music Comprehensive Understanding Evaluation). Experiments show our model significantly outperforms existing audio large language models across the MuCUE tasks, demonstrating its state-of-the-art effectiveness and generalization ability.

cs.SD↗

arXiv2025

Better Recommendations: Validating AI-generated Subject Terms Through LOC Linked Data Service

This article explores the integration of AI-generated subject terms into library cataloging, focusing on validation through the Library of Congress Linked Data Service. It examines the challenges of traditional subject cataloging under the Library of Congress Subject Headings system, including inefficiencies and cataloging backlogs. While generative AI shows promise in expediting cataloging workflows, studies reveal significant limitations in the accuracy of AI-assigned subject headings. The article proposes a hybrid approach combining AI technology with human validation through LOC Linked Data Service, aiming to enhance the precision, efficiency, and overall quality of metadata creation in library cataloging practices.

cs.DL↗

arXiv2025

Revealing electron-electron interactions in graphene at room temperature with the quantum twisting microscope

The Quantum Twisting Microscope (QTM) is a groundbreaking instrument that enables energy- and momentum-resolved measurements of quantum phases via tunneling spectroscopy across twistable van der Waals heterostructures. In this work, we significantly enhance the QTMs resolution and extend its measurement capabilities to higher energies and twist angles by incorporating hexagonal boron nitride (hBN) as a tunneling dielectric. This advancement unveils previously inaccessible signatures of the dispersion in the tunneling between two monolayer graphene (MLG) sheets, features consistent with a logarithmic correction to the linear Dirac dispersion arising from electron-electron (e-e) interactions with a fine-structure constant of alpha = 0.32. Remarkably, we find that this effect, for the first time, can be resolved even at room temperature, where these corrections are extremely faint. Our results underscore the exceptional resolution of the QTM, which, through interferometric interlayer tunneling, can amplify even subtle modifications to the electronic band structure of two-dimensional materials. Our findings reveal that strong e-e interactions persist even in symmetric, nonordered graphene states and emphasize the QTMs unique ability to probe spectral functions and their excitations of strongly correlated ground states across a broad range of twisted and untwisted systems.

cond-mat.mes-hall↗

arXiv2025

Quantum Geometry in the NbSe$_2$ Family I: Obstructed Compact Wannier Function and New Perturbation Theory

We revisit the electronic structure and band topology of monolayer 1H-NbSe$_2$, which hosts both superconductivity and charge density wave, and its related compounds 1H-MoS$_2$, NbS$_2$, TaS$_2$, TaSe$_2$ and WS$_2$. We construct a 6-band, a 3-band, and - simplest of all - a single-band model for this material family, by directly Wannierizing the ab initio bands. All host obstructed atomic isolated bands away from the atomic positions near the Fermi energy. We find that in the 3-band model, the obstructed atomic Wannier function can be well approximated by an optimally compact Wannier function with more than 90% accuracy for all the compounds, rising to a remarkable 94% accuracy in NbSe$_2$. Interestingly, the simplest single-band model has next nearest-neighboring hopping larger than the nearest-neighboring hopping (by nearly an order of magnitude for MoS$_2$, NbSe$_2$, TaSe$_2$ and WS$_2$), which comes from the cancellation between the atomic onsite terms and the atomic nearest-neighboring hopping after projecting to the obstructed atomic Wannier functions. Furthermore for NbSe$_2$, we employ a novel approximation scheme to obtain an effective Hamiltonian that captures the 3 bands originating mainly from the Nb atom. We also use conventional perturbation theory to derive the ab initio obstructed Wannier function with 95% accuracy. Our results pave the way for future study of the effect of quantum geometry on the correlated phases in this family of materials.

cond-mat.mes-hall↗