arXiv subjects

Li Zhang

Publications and source records attributed to Li Zhang.

At least 469 records · Page 26Linked to original sources

arXiv2024

FrameQuant: Flexible Low-Bit Quantization for Transformers

Transformers are the backbone of powerful foundation models for many Vision and Natural Language Processing tasks. But their compute and memory/storage footprint is large, and so, serving such models is expensive often requiring high-end hardware. To mitigate this difficulty, Post-Training Quantization seeks to modify a pre-trained model and quantize it to eight bits or lower, significantly boosting compute/memory/latency efficiency. Such models have been successfully quantized to four bits with some performance loss. In this work, we outline a simple scheme to quantize Transformer-based models to just two bits (plus some overhead) with only a small drop in accuracy. Key to our formulation is a concept borrowed from Harmonic analysis called Fusion Frames. Our main finding is that the quantization must take place not in the original weight space, but instead in the Fusion Frame representations. If quantization is interpreted as the addition of noise, our casting of the problem allows invoking an extensive body of known consistent recovery and noise robustness guarantees. Further, if desired, de-noising filters are known in closed form. We show empirically, via a variety of experiments, that (almost) two-bit quantization for Transformer models promises sizable efficiency gains. The code is available at https://github.com/vsingh-group/FrameQuant

cs.LG↗

arXiv2024

DuA: Dual Attentive Transformer in Long-Term Continuous EEG Emotion Analysis

Affective brain-computer interfaces (aBCIs) are increasingly recognized for their potential in monitoring and interpreting emotional states through electroencephalography (EEG) signals. Current EEG-based emotion recognition methods perform well with short segments of EEG data. However, these methods encounter significant challenges in real-life scenarios where emotional states evolve over extended periods. To address this issue, we propose a Dual Attentive (DuA) transformer framework for long-term continuous EEG emotion analysis. Unlike segment-based approaches, the DuA transformer processes an entire EEG trial as a whole, identifying emotions at the trial level, referred to as trial-based emotion analysis. This framework is designed to adapt to varying signal lengths, providing a substantial advantage over traditional methods. The DuA transformer incorporates three key modules: the spatial-spectral network module, the temporal network module, and the transfer learning module. The spatial-spectral network module simultaneously captures spatial and spectral information from EEG signals, while the temporal network module detects temporal dependencies within long-term EEG data. The transfer learning module enhances the model's adaptability across different subjects and conditions. We extensively evaluate the DuA transformer using a self-constructed long-term EEG emotion database, along with two benchmark EEG emotion databases. On the basis of the trial-based leave-one-subject-out cross-subject cross-validation protocol, our experimental results demonstrate that the proposed DuA transformer significantly outperforms existing methods in long-term continuous EEG emotion analysis, with an average enhancement of 5.28%.

cs.HC↗

arXiv2024

Reason2Drive: Towards Interpretable and Chain-based Reasoning for Autonomous Driving

Large vision-language models (VLMs) have garnered increasing interest in autonomous driving areas, due to their advanced capabilities in complex reasoning tasks essential for highly autonomous vehicle behavior. Despite their potential, research in autonomous systems is hindered by the lack of datasets with annotated reasoning chains that explain the decision-making processes in driving. To bridge this gap, we present Reason2Drive, a benchmark dataset with over 600K video-text pairs, aimed at facilitating the study of interpretable reasoning in complex driving environments. We distinctly characterize the autonomous driving process as a sequential combination of perception, prediction, and reasoning steps, and the question-answer pairs are automatically collected from a diverse range of open-source outdoor driving datasets, including nuScenes, Waymo and ONCE. Moreover, we introduce a novel aggregated evaluation metric to assess chain-based reasoning performance in autonomous systems, addressing the semantic ambiguities of existing metrics such as BLEU and CIDEr. Based on the proposed benchmark, we conduct experiments to assess various existing VLMs, revealing insights into their reasoning capabilities. Additionally, we develop an efficient approach to empower VLMs to leverage object-level perceptual elements in both feature extraction and prediction, further enhancing their reasoning accuracy. The code and dataset will be released.

cs.CV↗

arXiv2024

More is Different: Prototyping and Analyzing a New Form of Edge Server with Massive Mobile SoCs

Huge energy consumption poses a significant challenge for edge clouds. In response to this, we introduce a new type of edge server, namely SoC Cluster, that orchestrates multiple low-power mobile system-on-chips (SoCs) through an on-chip network. For the first time, we have developed a concrete SoC Cluster consisting of 60 Qualcomm Snapdragon 865 SoCs housed in a 2U rack, which has been successfully commercialized and extensively deployed in edge clouds. Cloud gaming emerges as the principal workload on these deployed SoC Clusters, owing to the compatibility between mobile SoCs and native mobile games. In this study, we aim to demystify whether the SoC Cluster can efficiently serve more generalized, typical edge workloads. Therefore, we developed a benchmark suite that employs state-of-the-art libraries for two critical edge workloads, i.e., video transcoding and deep learning inference. This suite evaluates throughput, latency, power consumption, and other application-specific metrics like video quality. Following this, we conducted a thorough measurement study and directly compared the SoC Cluster with traditional edge servers, with regards to electricity usage and monetary cost. Our results quantitatively reveal when and for which applications mobile SoCs exhibit higher energy efficiency than traditional servers, as well as their ability to proportionally scale power consumption with fluctuating incoming loads. These outcomes provide insightful implications and offer valuable direction for further refinement of the SoC Cluster to facilitate its deployment across wider edge scenarios.

cs.DC↗

arXiv2024

Uncovering Weaknesses in Neural Code Generation

Code generation, the task of producing source code from prompts, has seen significant advancements with the advent of pre-trained large language models (PLMs). Despite these achievements, there lacks a comprehensive taxonomy of weaknesses about the benchmark and the generated code, which risks the community's focus on known issues at the cost of under-explored areas. Our systematic study aims to fill this gap by evaluating five state-of-the-art PLMs: three larger models, CodeGen2.5 with 7 billion parameters, CodeGeeX2 with 6 billion parameters, GPT-4 Turbo, and two smaller ones, UnixCoder with 110 million parameters and CodeT5 base with 220 million parameters, across three popular datasets, CoNaLa, HumanEval Plus, and DS-1000. We assess the quality of generated code using match-based and execution-based metrics, then conduct thematic analysis to develop a taxonomy of nine types of weaknesses. We dissected weakness distributions in both larger and smaller models, applying an extensive methodology that encompasses model-specific as well as collective analysis (union and intersection) across models. Our research uncovers three salient findings: 1. In the CoNaLa dataset, inaccurate prompts are a notable problem, causing all large models to fail in 26.84% of cases, with even higher failure rates of 40% for smaller models; 2. Missing pivotal semantics is a pervasive issue across benchmarks, with one or more large models omitting key semantics in 65.78% of CoNaLa tasks, and similarly high occurrences in HumanEval Plus (66.09%) and DS-1000 (80.51%); 3. All models struggle with proper API usage, a challenge amplified by vague or complex prompts. Our findings aim to steer researchers towards addressing specific weaknesses and challenges in code generation. Furthermore, our annotations can offer a targeted benchmark subset for detailed analysis.

cs.SE↗

arXiv2024

MoE-DiffIR: Task-customized Diffusion Priors for Universal Compressed Image Restoration

We present MoE-DiffIR, an innovative universal compressed image restoration (CIR) method with task-customized diffusion priors. This intends to handle two pivotal challenges in the existing CIR methods: (i) lacking adaptability and universality for different image codecs, e.g., JPEG and WebP; (ii) poor texture generation capability, particularly at low bitrates. Specifically, our MoE-DiffIR develops the powerful mixture-of-experts (MoE) prompt module, where some basic prompts cooperate to excavate the task-customized diffusion priors from Stable Diffusion (SD) for each compression task. Moreover, the degradation-aware routing mechanism is proposed to enable the flexible assignment of basic prompts. To activate and reuse the cross-modality generation prior of SD, we design the visual-to-text adapter for MoE-DiffIR, which aims to adapt the embedding of low-quality images from the visual domain to the textual domain as the textual guidance for SD, enabling more consistent and reasonable texture generation. We also construct one comprehensive benchmark dataset for universal CIR, covering 21 types of degradations from 7 popular traditional and learned codecs. Extensive experiments on universal CIR have demonstrated the excellent robustness and texture restoration capability of our proposed MoE-DiffIR. The project can be found at https://renyulin-f.github.io/MoE-DiffIR.github.io/.

eess.IV↗

arXiv2024

Whisper-SV: Adapting Whisper for Low-data-resource Speaker Verification

Trained on 680,000 hours of massive speech data, Whisper is a multitasking, multilingual speech foundation model demonstrating superior performance in automatic speech recognition, translation, and language identification. However, its applicability in speaker verification (SV) tasks remains unexplored, particularly in low-data-resource scenarios where labeled speaker data in specific domains are limited. To fill this gap, we propose a lightweight adaptor framework to boost SV with Whisper, namely Whisper-SV. Given that Whisper is not specifically optimized for SV tasks, we introduce a representation selection module to quantify the speaker-specific characteristics contained in each layer of Whisper and select the top-k layers with prominent discriminative speaker features. To aggregate pivotal speaker-related features while diminishing non-speaker redundancies across the selected top-k distinct layers of Whisper, we design a multi-layer aggregation module in Whisper-SV to integrate multi-layer representations into a singular, compacted representation for SV. In the multi-layer aggregation module, we employ convolutional layers with shortcut connections among different layers to refine speaker characteristics derived from multi-layer representations from Whisper. In addition, an attention aggregation layer is used to reduce non-speaker interference and amplify speaker-specific cues for SV tasks. Finally, a simple classification module is used for speaker classification. Experiments on VoxCeleb1, FFSVC, and IMSV datasets demonstrate that Whisper-SV achieves EER/minDCF of 2.22%/0.307, 6.14%/0.488, and 7.50%/0.582, respectively, showing superior performance in low-data-resource SV scenarios.

cs.SD↗

arXiv2024

PROC2PDDL: Open-Domain Planning Representations from Texts

Planning in a text-based environment continues to be a major challenge for AI systems. Recent approaches have used language models to predict a planning domain definition (e.g., PDDL) but have only been evaluated in closed-domain simulated environments. To address this, we present Proc2PDDL , the first dataset containing open-domain procedural texts paired with expert-annotated PDDL representations. Using this dataset, we evaluate state-of-the-art models on defining the preconditions and effects of actions. We show that Proc2PDDL is highly challenging, with GPT-3.5's success rate close to 0% and GPT-4's around 35%. Our analysis shows both syntactic and semantic errors, indicating LMs' deficiency in both generating domain-specific prgorams and reasoning about events. We hope this analysis and dataset helps future progress towards integrating the best of LMs and formal planning.

cs.CL↗

arXiv2024

RoDyn-SLAM: Robust Dynamic Dense RGB-D SLAM with Neural Radiance Fields

Leveraging neural implicit representation to conduct dense RGB-D SLAM has been studied in recent years. However, this approach relies on a static environment assumption and does not work robustly within a dynamic environment due to the inconsistent observation of geometry and photometry. To address the challenges presented in dynamic environments, we propose a novel dynamic SLAM framework with neural radiance field. Specifically, we introduce a motion mask generation method to filter out the invalid sampled rays. This design effectively fuses the optical flow mask and semantic mask to enhance the precision of motion mask. To further improve the accuracy of pose estimation, we have designed a divide-and-conquer pose optimization algorithm that distinguishes between keyframes and non-keyframes. The proposed edge warp loss can effectively enhance the geometry constraints between adjacent frames. Extensive experiments are conducted on the two challenging datasets, and the results show that RoDyn-SLAM achieves state-of-the-art performance among recent neural RGB-D methods in both accuracy and robustness.

cs.RO↗

arXiv2024

SCDNet: Self-supervised Learning Feature-based Speaker Change Detection

Speaker Change Detection (SCD) is to identify boundaries among speakers in a conversation. Motivated by the success of fine-tuning wav2vec 2.0 models for the SCD task, a further investigation of self-supervised learning (SSL) features for SCD is conducted in this work. Specifically, an SCD model, named SCDNet, is proposed. With this model, various state-of-the-art SSL models, including Hubert, wav2vec 2.0, and WavLm are investigated. To discern the most potent layer of SSL models for SCD, a learnable weighting method is employed to analyze the effectiveness of intermediate representations. Additionally, a fine-tuning-based approach is also implemented to further compare the characteristics of SSL models in the SCD task. Furthermore, a contrastive learning method is proposed to mitigate the overfitting tendencies in the training of both the fine-tuning-based method and SCDNet. Experiments showcase the superiority of WavLm in the SCD task and also demonstrate the good design of SCDNet.

eess.AS↗

arXiv2024

TimeSiam: A Pre-Training Framework for Siamese Time-Series Modeling

Time series pre-training has recently garnered wide attention for its potential to reduce labeling expenses and benefit various downstream tasks. Prior methods are mainly based on pre-training techniques well-acknowledged in vision or language, such as masked modeling and contrastive learning. However, randomly masking time series or calculating series-wise similarity will distort or neglect inherent temporal correlations crucial in time series data. To emphasize temporal correlation modeling, this paper proposes TimeSiam as a simple but effective self-supervised pre-training framework for Time series based on Siamese networks. Concretely, TimeSiam pre-trains Siamese encoders to capture intrinsic temporal correlations between randomly sampled past and current subseries. With a simple data augmentation method (e.g.~masking), TimeSiam can benefit from diverse augmented subseries and learn internal time-dependent representations through a past-to-current reconstruction. Moreover, learnable lineage embeddings are also introduced to distinguish temporal distance between sampled series and further foster the learning of diverse temporal correlations. TimeSiam consistently outperforms extensive advanced pre-training baselines, demonstrating superior forecasting and classification capabilities across 13 standard benchmarks in both intra- and cross-domain scenarios.

cs.LG↗

arXiv2024

Inpainting Pathology in Lumbar Spine MRI with Latent Diffusion

Data driven models for automated diagnosis in radiology suffer from insufficient and imbalanced datasets due to low representation of pathology in a population and the cost of expert annotations. Datasets can be bolstered through data augmentation. However, even when utilizing a full suite of transformations during model training, typical data augmentations do not address variations in human anatomy. An alternative direction is to synthesize data using generative models, which can potentially craft datasets with specific attributes. While this holds promise, commonly used generative models such as Generative Adversarial Networks may inadvertently produce anatomically inaccurate features. On the other hand, diffusion models, which offer greater stability, tend to memorize training data, raising concerns about privacy and generative diversity. Alternatively, inpainting has the potential to augment data through directly inserting pathology in medical images. However, this approach introduces a new challenge: accurately merging the generated pathological features with the surrounding anatomical context. While inpainting is a well established method for addressing simple lesions, its application to pathologies that involve complex structural changes remains relatively unexplored. We propose an efficient method for inpainting pathological features onto healthy anatomy in MRI through voxelwise noise scheduling in a latent diffusion model. We evaluate the method's ability to insert disc herniation and central canal stenosis in lumbar spine sagittal T2 MRI, and it achieves superior Frechet Inception Distance compared to state-of-the-art methods.

eess.IV↗

arXiv2024

Tailoring with Targeted Precision: Edit-Based Agents for Open-Domain Procedure Customization

How-to procedures, such as how to plant a garden, are now used by millions of users, but sometimes need customizing to meet a user's specific needs, e.g., planting a garden without pesticides. Our goal is to measure and improve an LLM's ability to perform such customization. Our approach is to test several simple multi-LLM-agent architectures for customization, as well as an end-to-end LLM, using a new evaluation set, called CustomPlans, of over 200 WikiHow procedures each with a customization need. We find that a simple architecture with two LLM agents used sequentially performs best, one that edits a generic how-to procedure and one that verifies its executability, significantly outperforming (10.5% absolute) an end-to-end prompted LLM. This suggests that LLMs can be configured reasonably effectively for procedure customization. This also suggests that multi-agent editing architectures may be worth exploring further for other customization applications (e.g. coding, creative writing) in the future.

cs.CL↗

arXiv2024

A Novel Approach for Automated Design Information Mining from Issue Logs

Software architectures are usually meticulously designed to address multiple quality concerns and support long-term maintenance. However, due to the imbalance between the cost and value for developers to document design rationales (i.e., the design alternatives and the underlying arguments for making or rejecting decisions), these rationales are often obsolete or even missing. The lack of design knowledge has motivated a number of studies to extract design information from various platforms in recent years. Unfortunately, despite the wealth of discussion records related to design information provided by platforms like open-source communities, existing research often overlooks the underlying arguments behind alternatives due to challenges such as the intricate semantics of discussions and the lack of benchmarks for design rationale extraction. In this paper, we propose a novel method, named by DRMiner, to automatically mine latent design rationales from developers' live discussion in open-source community (i.e., issue logs in Jira). To better identify solutions and the arguments supporting them, DRMiner skillfully decomposes the problem into multiple text classification tasks and tackles them using prompt tuning of language models and customized text-related features. To evaluate DRMiner, we acquire issue logs from Cassandra, Flink, and Solr repositories in Jira, and then annotate and process them under a rigorous scheme, ultimately forming a dataset for design rationale mining. Experimental results show that DRMiner achieves an F1 score of 65% for mining design rationales, outperforming all baselines with a 7% improvement over GPT-4.0. Furthermore, we investigate the usefulness of the design rationales mined by DRMiner for automated program repair (APR) and find that the design rationales significantly enhance APR, achieving 14 times higher full-match repairs on average.

cs.SE↗

arXiv2024

OpFlowTalker: Realistic and Natural Talking Face Generation via Optical Flow Guidance

Creating realistic, natural, and lip-readable talking face videos remains a formidable challenge. Previous research primarily concentrated on generating and aligning single-frame images while overlooking the smoothness of frame-to-frame transitions and temporal dependencies. This often compromised visual quality and effects in practical settings, particularly when handling complex facial data and audio content, which frequently led to semantically incongruent visual illusions. Specifically, synthesized videos commonly featured disorganized lip movements, making them difficult to understand and recognize. To overcome these limitations, this paper introduces the application of optical flow to guide facial image generation, enhancing inter-frame continuity and semantic consistency. We propose "OpFlowTalker", a novel approach that utilizes predicted optical flow changes from audio inputs rather than direct image predictions. This method smooths image transitions and aligns changes with semantic content. Moreover, it employs a sequence fusion technique to replace the independent generation of single frames, thus preserving contextual information and maintaining temporal coherence. We also developed an optical flow synchronization module that regulates both full-face and lip movements, optimizing visual synthesis by balancing regional dynamics. Furthermore, we introduce a Visual Text Consistency Score (VTCS) that accurately measures lip-readability in synthesized videos. Extensive empirical evidence validates the effectiveness of our approach.

cs.CV↗

arXiv2024

"Pass the butter": A study on desktop-classic multitasking robotic arm based on advanced YOLOv7 and BERT

In recent years, various intelligent autonomous robots have begun to appear in daily life and production. Desktop-level robots are characterized by their flexible deployment, rapid response, and suitability for light workload environments. In order to meet the current societal demand for service robot technology, this study proposes using a miniaturized desktop-level robot (by ROS) as a carrier, locally deploying a natural language model (NLP-BERT), and integrating visual recognition (CV-YOLO) and speech recognition technology (ASR-Whisper) as inputs to achieve autonomous decision-making and rational action by the desktop robot. Three comprehensive experiments were designed to validate the robotic arm, and the results demonstrate excellent performance using this approach across all three experiments. In Task 1, the execution rates for speech recognition and action performance were 92.6% and 84.3%, respectively. In Task 2, the highest execution rates under the given conditions reached 92.1% and 84.6%, while in Task 3, the highest execution rates were 95.2% and 80.8%, respectively. Therefore, it can be concluded that the proposed solution integrating ASR, NLP, and other technologies on edge devices is feasible and provides a technical and engineering foundation for realizing multimodal desktop-level robots.

cs.RO↗

arXiv2024

Measurement Embedded Schrödinger Bridge for Inverse Problems

Score-based diffusion models are frequently employed as structural priors in inverse problems. However, their iterative denoising process, initiated from Gaussian noise, often results in slow inference speeds. The Image-to-Image Schrödinger Bridge (I$^2$SB), which begins with the corrupted image, presents a promising alternative as a prior for addressing inverse problems. In this work, we introduce the Measurement Embedded Schrödinger Bridge (MESB). MESB establishes Schrödinger Bridges between the distribution of corrupted images and the distribution of clean images given observed measurements. Based on optimal transport theory, we derive the forward and backward processes of MESB. Through validation on diverse inverse problems, our proposed approach exhibits superior performance compared to existing Schrödinger Bridge-based inverse problems solvers in both visual quality and quantitative metrics.

eess.IV↗

arXiv2024

Survey on Visual Signal Coding and Processing with Generative Models: Technologies, Standards and Optimization

This paper provides a survey of the latest developments in visual signal coding and processing with generative models. Specifically, our focus is on presenting the advancement of generative models and their influence on research in the domain of visual signal coding and processing. This survey study begins with a brief introduction of well-established generative models, including the Variational Autoencoder (VAE) models, Generative Adversarial Network (GAN) models, Autoregressive (AR) models, Normalizing Flows and Diffusion models. The subsequent section of the paper explores the advancements in visual signal coding based on generative models, as well as the ongoing international standardization activities. In the realm of visual signal processing, our focus lies on the application and development of various generative models in the research of visual signal restoration. We also present the latest developments in generative visual signal synthesis and editing, along with visual signal quality assessment using generative models and quality assessment for generative models. The practical implementation of these studies is closely linked to the investigation of fast optimization. This paper additionally presents the latest advancements in fast optimization on visual signal coding and processing with generative models. We hope to advance this field by providing researchers and practitioners a comprehensive literature review on the topic of visual signal coding and processing with generative models.

eess.IV↗