Search arXiv⌕ Search

arXiv subjects

Yu Zhang

Publications and source records attributed to Yu Zhang.

At least 721 records · Page 40Linked to original sources
arXiv2024

Gated Slot Attention for Efficient Linear-Time Sequence Modeling

Linear attention Transformers and their gated variants, celebrated for enabling parallel training and efficient recurrent inference, still fall short in recall-intensive tasks compared to traditional Transformers and demand significant resources for training from scratch. This paper introduces Gated Slot Attention (GSA), which enhances Attention with Bounded-memory-Control (ABC) by incorporating a gating mechanism inspired by Gated Linear Attention (GLA). Essentially, GSA comprises a two-layer GLA linked via $\operatorname{softmax}$, utilizing context-aware memory reading and adaptive forgetting to improve memory capacity while maintaining compact recurrent state size. This design greatly enhances both training and inference efficiency through GLA's hardware-efficient training algorithm and reduced state size. Additionally, retaining the $\operatorname{softmax}$ operation is particularly beneficial in "finetuning pretrained Transformers to RNNs" (T2R) settings, reducing the need for extensive training from scratch. Extensive experiments confirm GSA's superior performance in scenarios requiring in-context recall and in T2R settings.

cs.CL↗
arXiv2024

Multi-Task Interactive Robot Fleet Learning with Visual World Models

Recent advancements in large-scale multi-task robot learning offer the potential for deploying robot fleets in household and industrial settings, enabling them to perform diverse tasks across various environments. However, AI-enabled robots often face challenges with generalization and robustness when exposed to real-world variability and uncertainty. We introduce Sirius-Fleet, a multi-task interactive robot fleet learning framework to address these challenges. Sirius-Fleet monitors robot performance during deployment and involves humans to correct the robot's actions when necessary. We employ a visual world model to predict the outcomes of future actions and build anomaly predictors to predict whether they will likely result in anomalies. As the robot autonomy improves, the anomaly predictors automatically adapt their prediction criteria, leading to fewer requests for human intervention and gradually reducing human workload over time. Evaluations on large-scale benchmarks demonstrate Sirius-Fleet's effectiveness in improving multi-task policy performance and monitoring accuracy. We demonstrate Sirius-Fleet's performance in both RoboCasa in simulation and Mutex in the real world, two diverse, large-scale multi-task benchmarks. More information is available on the project website: https://ut-austin-rpl.github.io/sirius-fleet

cs.RO↗
arXiv2024

CopRA: A Progressive LoRA Training Strategy

Low-Rank Adaptation (LoRA) is a parameter-efficient technique for rapidly fine-tuning foundation models. In standard LoRA training dynamics, models tend to quickly converge to a local optimum near the initialization. However, this local optimum may not be ideal for out-of-distribution data or tasks such as merging and pruning. In this work, we propose a novel progressive training strategy for LoRA with random layer dropping. This strategy also optimizes the Shapley value of LoRA parameters in each layer, treating each layer as a player in a cooperative game. We refer to this method as Cooperative LoRA (CopRA). Our experimental results demonstrate that parameters trained with CopRA exhibit linear mode connectivity, which enables efficient model merging. This also paves the way for federated learning and multi-task learning via LoRA merging. Additionally, by optimizing the Shapley value, CopRA shows superior performance in pruning tasks.

cs.LG↗
arXiv2024

Quantifying the non-Abelian property of Andreev bound states in inhomogeneous Majorana nanowires

Non-Abelian braiding is a key property of Majorana zero modes (MZMs) that can be utilized for topological quantum computation. However, the presence of trivial Andreev bound states (ABSs) in topological superconductors can hinder the non-Abelian braiding of MZMs. We systematically investigate the braiding properties of ABSs induced by various inhomogeneous potentials in nanowires and quantify the main obstacles to non-Abelian braiding. We find that if a trivial ABSs is present at zero energy with a tiny energy fluctuation, their non-Abelian braiding property can be sustained for a longer braiding time cost, since the undesired dynamic phase is suppressed. Under certain conditions, the non-Abelian braiding of ABSs can even surpass that of MZMs in realistic systems, suggesting that ABSs might also be suitable for topological quantum computation.

cond-mat.mes-hall↗
arXiv2024

MSGField: A Unified Scene Representation Integrating Motion, Semantics, and Geometry for Robotic Manipulation

Combining accurate geometry with rich semantics has been proven to be highly effective for language-guided robotic manipulation. Existing methods for dynamic scenes either fail to update in real-time or rely on additional depth sensors for simple scene editing, limiting their applicability in real-world. In this paper, we introduce MSGField, a representation that uses a collection of 2D Gaussians for high-quality reconstruction, further enhanced with attributes to encode semantic and motion information. Specially, we represent the motion field compactly by decomposing each primitive's motion into a combination of a limited set of motion bases. Leveraging the differentiable real-time rendering of Gaussian splatting, we can quickly optimize object motion, even for complex non-rigid motions, with image supervision from only two camera views. Additionally, we designed a pipeline that utilizes object priors to efficiently obtain well-defined semantics. In our challenging dataset, which includes flexible and extremely small objects, our method achieve a success rate of 79.2% in static and 63.3% in dynamic environments for language-guided manipulation. For specified object grasping, we achieve a success rate of 90%, on par with point cloud-based methods. Code and dataset will be released at:https://shengyu724.github.io/MSGField.github.io.

cs.RO↗
arXiv2024

Semiclassical Nonadiabatic Molecular Dynamics for Molecular Exciton-Polaritons

When the interaction between a molecular system and confined light modes in an optical or plasmonic cavity is strong enough to overcome the dissipative process, hybrid light-matter states (polaritons) emerge as the fundamental excitations in the system. Mixing the light and matter characters modifies molecules' photophysical and photochemical properties. It was reported that polaritonic states can be employed to control photochemical reactions, charge and energy transfer, and other processes. In addition, according to recent studies, vibrational strong coupling can be employed to enhance thermally activated chemical reactions resonantly. This work adopts a coherent state-based many-body state as the basis function to expand the light-matter Hamiltonian. The corresponding nonadiabatic Molecular Dynamics scheme is derived and implemented in the NEXMD package based on the semiclassical Ab Initio Multiple Cloning (AIMC) protocol. The scheme is demonstrated via a model system, pyridine, to demonstrate its validity. Our numerical results show that coherence and decoherence processes are distinguishable during the initial relaxation time in the AIMC simulations, while such dynamics are missing in the mixed quantum-classical surface hopping approach. Additionally, our results show that the light-matter coupling has smaller impact on the dynamics in surface hopping simulations. In the AIMC simulations, on the other hand, the relaxation time is increased, leading to a slower internal conversion and decreased equilibrium population build-up in the lowest electronically excited state. This highlights the importance of proper treatment of coherence in simulating exciton polaritons dynamics.

physics.chem-ph↗
arXiv2024

Structural and electronic transformations in TiO2 induced by electric current

In-situ diffuse neutron scattering experiments revealed that when electric current is passed through single crystals of rutile TiO2 under conditions conducive to flash sintering, it induces the formation of parallel planes of oxygen vacancies. Specifically, a current perpendicular to the c-axis generates planes normal to the (132) reciprocal lattice vector, whereas currents aligned with the c-axis form planes normal to the (132) and to the (225) vector. The concentration of defects increases with incresing current. The structural modifications are linked to the appearance of signatures of interacting Ti3+ moments in magnetic susceptibility, signifying a structural collapse around the vacancy planes. Electrical conductivity measurements of the modified material reveal several electronic transitions between semiconducting states (via a metal-like intermediate state) with the smallest gap being 27 meV. Pristine TiO2 can be restored by heating followed by slow cooling in air. Our work suggests a novel paradigm for achieving switching of electrical conductivity related to the flash phenomenon

cond-mat.mtrl-sci↗
arXiv2024

Ultrafast Processes in 1,2-Dichloroethene measured with a Universal XUV probe

The presence of two chlorine atoms in 1,2-dichloroethene allows for isomerisation around the double bond. This isomerisation can lead to rich photochemistry. We present a time-resolved pump-probe photoelectron spectroscopy measurement on both the cis- and trans- isomers of 1,2-dichloroethene. A universal XUV probe of 22.3 eV is used allowing observation of photoelectrons formed anywhere on the potential energy surface, even from the ground-state or dissociation products. Following excitation with a 200 nm probe both ultrafast excited state dynamics and product formation are observed within the time resolution of the experiment. Excited state population begins to return to the ground state on an ultrafast time scale (< 70 fs) and population of products channels is observed on the same timescale. With the aid of ab initio calculations it is found that population transfer from the excited state is facilitated by vibrational modes involving C-C-H bends.

physics.chem-ph↗
arXiv2024

Sharpness-Aware Black-Box Optimization

Black-box optimization algorithms have been widely used in various machine learning problems, including reinforcement learning and prompt fine-tuning. However, directly optimizing the training loss value, as commonly done in existing black-box optimization methods, could lead to suboptimal model quality and generalization performance. To address those problems in black-box optimization, we propose a novel Sharpness-Aware Black-box Optimization (SABO) algorithm, which applies a sharpness-aware minimization strategy to improve the model generalization. Specifically, the proposed SABO method first reparameterizes the objective function by its expectation over a Gaussian distribution. Then it iteratively updates the parameterized distribution by approximated stochastic gradients of the maximum objective value within a small neighborhood around the current solution in the Gaussian distribution space. Theoretically, we prove the convergence rate and generalization bound of the proposed SABO algorithm. Empirically, extensive experiments on the black-box prompt fine-tuning tasks demonstrate the effectiveness of the proposed SABO method in improving model generalization performance.

cs.LG↗
arXiv2024

Eyes Closed, Safety On: Protecting Multimodal LLMs via Image-to-Text Transformation

Multimodal large language models (MLLMs) have shown impressive reasoning abilities. However, they are also more vulnerable to jailbreak attacks than their LLM predecessors. Although still capable of detecting the unsafe responses, we observe that safety mechanisms of the pre-aligned LLMs in MLLMs can be easily bypassed with the introduction of image features. To construct robust MLLMs, we propose ECSO (Eyes Closed, Safety On), a novel training-free protecting approach that exploits the inherent safety awareness of MLLMs, and generates safer responses via adaptively transforming unsafe images into texts to activate the intrinsic safety mechanism of pre-aligned LLMs in MLLMs. Experiments on five state-of-the-art (SoTA) MLLMs demonstrate that ECSO enhances model safety significantly (e.g.,, 37.6% improvement on the MM-SafetyBench (SD+OCR) and 71.3% on VLSafe with LLaVA-1.5-7B), while consistently maintaining utility results on common MLLM benchmarks. Furthermore, we show that ECSO can be used as a data engine to generate supervised-finetuning (SFT) data for MLLM alignment without extra human intervention.

cs.CV↗
arXiv2024

AutoTurb: Using Large Language Models for Automatic Algebraic Model Discovery of Turbulence Closure

Symbolic regression (SR) methods have been extensively investigated to explore explicit algebraic Reynolds stress models (EARSM) for turbulence closure of Reynolds-averaged Navier-Stokes (RANS) equations. The deduced EARSM can be readily implemented in existing computational fluid dynamic (CFD) codes and promotes the identification of physically interpretable turbulence models. The existing SR methods, such as genetic programming, sparse regression, or artificial neural networks, require user-defined functional operators, a library of candidates, or complex optimization algorithms. In this work, a novel framework using LLMs to automatically discover algebraic expressions for correcting the RSM is proposed. The direct observation of Reynolds stress and the indirect output of the CFD simulation are both involved in the training process to guarantee data consistency and avoid numerical stiffness. Constraints of functional complexity and convergence are supplementally imposed in the objective function on account of the tremendous flexibility of LLMs. The evolutionary search is employed for global optimization. The proposed method is performed for separated flow over periodic hills at Re = 10,595. The generalizability of the discovered model is verified on a set of 2D turbulent separated flow configurations with different Reynolds numbers and geometries. It is demonstrated that the corrective RANS can improve the prediction for both the Reynolds stress and mean velocity fields. Compared with algebraic models discovered by other works, the discovered model performs better in accuracy and generalization capability. The proposed approach provides a promising paradigm for using LLMs to improve turbulence modeling for a given class of flows.

physics.flu-dyn↗
arXiv2024

DAT: Dialogue-Aware Transformer with Modality-Group Fusion for Human Engagement Estimation

Engagement estimation plays a crucial role in understanding human social behaviors, attracting increasing research interests in fields such as affective computing and human-computer interaction. In this paper, we propose a Dialogue-Aware Transformer framework (DAT) with Modality-Group Fusion (MGF), which relies solely on audio-visual input and is language-independent, for estimating human engagement in conversations. Specifically, our method employs a modality-group fusion strategy that independently fuses audio and visual features within each modality for each person before inferring the entire audio-visual content. This strategy significantly enhances the model's performance and robustness. Additionally, to better estimate the target participant's engagement levels, the introduced Dialogue-Aware Transformer considers both the participant's behavior and cues from their conversational partners. Our method was rigorously tested in the Multi-Domain Engagement Estimation Challenge held by MultiMediate'24, demonstrating notable improvements in engagement-level regression precision over the baseline model. Notably, our approach achieves a CCC score of 0.76 on the NoXi Base test set and an average CCC of 0.64 across the NoXi Base, NoXi-Add, and MPIIGI test sets.

cs.HC↗
arXiv2024

A Lightweight Target-Driven Network of Stereo Matching for Inland Waterways

Stereo matching for inland waterways is one of the key technologies for the autonomous navigation of Unmanned Surface Vehicles (USVs), which involves dividing the stereo images into reference images and target images for pixel-level matching. However, due to the challenges of the inland waterway environment, such as blurred textures, large spatial scales, and computational resource constraints of the USVs platform, the participation of geometric features from the target image is required for efficient target-driven matching. Based on this target-driven concept, we propose a lightweight target-driven stereo matching neural network, named LTNet. Specifically, a lightweight and efficient 4D cost volume, named the Geometry Target Volume (GTV), is designed to fully utilize the geometric information of target features by employing the shifted target features as the filtered feature volume. Subsequently, to address the substantial texture interference and object occlusions present in the waterway environment, a Left-Right Consistency Refinement (LRR) module is proposed. The \text{LRR} utilizes the pixel-level differences in left and right disparities to introduce soft constraints, thereby enhancing the accuracy of predictions during the intermediate stages of the network. Moreover, knowledge distillation is utilized to enhance the generalization capability of lightweight models on the USVInland dataset. Furthermore, a new large-scale benchmark, named Spring, is utilized to validate the applicability of LTNet across various scenarios. In experiments on the aforementioned two datasets, LTNet achieves competitive results, with only 3.7M parameters. The code is available at https://github.com/Open-YiQingZhou/LTNet .

cs.CV↗
arXiv2024

Improved PCRLB for radar tracking in clutter with geometry-dependent target measurement uncertainty and application to radar trajectory control

In realistic radar tracking, target measurement uncertainty (TMU) in terms of both detection probability and measurement error covariance is significantly affected by the target-to-radar (T2R) geometry. However, existing posterior Cramer-Rao Lower Bounds (PCRLBs) rarely investigate the fundamental impact of T2R geometry on target measurement uncertainty and eventually on mean square error (MSE) of state estimate, inevitably resulting in over-conservative lower bound. To address this issue, this paper firstly derives the generalized model of target measurement error covariance for bistatic radar with moving receiver and transmitter illuminating any type of signal, along with its approximated solution to specify the impact of T2R geometry on error covariance. Based upon formulated TMU model, an improved PCRLB (IPCRLB) fully accounting for both measurement origin uncertainty and geometry-dependent TMU is then re-derived, both detection probability and measurement error covariance are treated as state-dependent parameters when differentiating log-likelihood with respect to target state. Compared to existing PCRLBs that partially or completely ignore the dependence of target measurement uncertainty on T2R geometry, proposed IPCRLB provides a much accurate (less-conservative) lower bound for radar tracking in clutter with geometry-dependent TMU. The new bound is then applied to radar trajectory control to effectively optimize T2R geometry and exhibits least uncertainty of acquired target measurement and more accurate state estimate for bistatic radar tracking in clutter, compared to state-of-the-art trajectory control methods.

eess.SP↗
arXiv2024

IDGen: Item Discrimination Induced Prompt Generation for LLM Evaluation

As Large Language Models (LLMs) grow increasingly adept at managing complex tasks, the evaluation set must keep pace with these advancements to ensure it remains sufficiently discriminative. Item Discrimination (ID) theory, which is widely used in educational assessment, measures the ability of individual test items to differentiate between high and low performers. Inspired by this theory, we propose an ID-induced prompt synthesis framework for evaluating LLMs to ensure the evaluation set can continually update and refine according to model abilities. Our data synthesis framework prioritizes both breadth and specificity. It can generate prompts that comprehensively evaluate the capabilities of LLMs while revealing meaningful performance differences between models, allowing for effective discrimination of their relative strengths and weaknesses across various tasks and domains. To produce high-quality data, we incorporate a self-correct mechanism into our generalization framework, and develop two models to predict prompt discrimination and difficulty score to facilitate our data synthesis framework, contributing valuable tools to evaluation data synthesis research. We apply our generated data to evaluate five SOTA models. Our data achieves an average score of 51.92, accompanied by a variance of 10.06. By contrast, previous works (i.e., SELF-INSTRUCT and WizardLM) obtain an average score exceeding 67, with a variance below 3.2. The results demonstrate that the data generated by our framework is more challenging and discriminative compared to previous works. We will release a dataset of over 3,000 carefully crafted prompts to facilitate evaluation research of LLMs.

cs.CL↗
arXiv2024

Graph Chain-of-Thought: Augmenting Large Language Models by Reasoning on Graphs

Large language models (LLMs), while exhibiting exceptional performance, suffer from hallucinations, especially on knowledge-intensive tasks. Existing works propose to augment LLMs with individual text units retrieved from external knowledge corpora to alleviate the issue. However, in many domains, texts are interconnected (e.g., academic papers in a bibliographic graph are linked by citations and co-authorships) which form a (text-attributed) graph. The knowledge in such graphs is encoded not only in single texts/nodes but also in their associated connections. To facilitate the research of augmenting LLMs with graphs, we manually construct a Graph Reasoning Benchmark dataset called GRBench, containing 1,740 questions that can be answered with the knowledge from 10 domain graphs. Then, we propose a simple and effective framework called Graph Chain-of-thought (Graph-CoT) to augment LLMs with graphs by encouraging LLMs to reason on the graph iteratively. Each Graph-CoT iteration consists of three sub-steps: LLM reasoning, LLM-graph interaction, and graph execution. We conduct systematic experiments with three LLM backbones on GRBench, where Graph-CoT outperforms the baselines consistently. The code is available at https://github.com/PeterGriffinJin/Graph-CoT.

cs.CL↗
arXiv2024

Question-guided Knowledge Graph Re-scoring and Injection for Knowledge Graph Question Answering

Knowledge graph question answering (KGQA) involves answering natural language questions by leveraging structured information stored in a knowledge graph. Typically, KGQA initially retrieve a targeted subgraph from a large-scale knowledge graph, which serves as the basis for reasoning models to address queries. However, the retrieved subgraph inevitably brings distraction information for knowledge utilization, impeding the model's ability to perform accurate reasoning. To address this issue, we propose a Question-guided Knowledge Graph Re-scoring method (Q-KGR) to eliminate noisy pathways for the input question, thereby focusing specifically on pertinent factual knowledge. Moreover, we introduce Knowformer, a parameter-efficient method for injecting the re-scored knowledge graph into large language models to enhance their ability to perform factual reasoning. Extensive experiments on multiple KGQA benchmarks demonstrate the superiority of our method over existing systems.

cs.CL↗