arXiv subjects

Xiaochen Li

Publications and source records attributed to Xiaochen Li.

At least 37 records · Page 2Linked to original sources

arXiv2025

A Novel Mutation Based Method for Detecting FPGA Logic Synthesis Tool Bugs

FPGA (Field-Programmable Gate Array) logic synthesis tools are key components in the EDA (Electronic Design Automation) toolchain. They convert hardware designs written in description languages such as Verilog into gate-level representations for FPGAs. However, defects in these tools may lead to unexpected behaviors and pose security risks. Therefore, it is crucial to harden these tools through testing. Although several methods have been proposed to automatically test FPGA logic synthesis tools, the challenge remains of insufficient semantic and logical complexity in test programs. In this paper, we propose VERMEI, a new method for testing FPGA logic synthesis tools. VERMEI consists of three modules: preprocessing, equivalent mutation, and bug identification. The preprocessing module identifies zombie logic (inactive code with no impact on the circuit output) in seed programs through simulation and coverage analysis. The equivalent mutation module generates equivalent variants of seed programs by pruning or inserting logic fragments in zombie areas. It uses Bayesian sampling to extract logic fragments from historical Verilog designs, making the generated variants have complex control flows and structures. The bug identification module, based on differential testing, compares the synthesized outputs of seed and variant programs to identify bugs. Experiments on Yosys, Vivado, and Quartus demonstrate that VERMEI outperforms the state-of-the-art methods. Within five months, VERMEI reported 15 bugs to vendors, 9 of which were confirmed as new.

cs.SE↗

arXiv2025

Towards Evaluation for Real-World LLM Unlearning

This paper analyzes the limitations of existing unlearning evaluation metrics in terms of practicality, exactness, and robustness in real-world LLM unlearning scenarios. To overcome these limitations, we propose a new metric called Distribution Correction-based Unlearning Evaluation (DCUE). It identifies core tokens and corrects distributional biases in their confidence scores using a validation set. The evaluation results are quantified using the Kolmogorov-Smirnov test. Experimental results demonstrate that DCUE overcomes the limitations of existing metrics, which also guides the design of more practical and reliable unlearning algorithms in the future.

cs.AI↗

arXiv2025

From Easy to Hard: Building a Shortcut for Differentially Private Image Synthesis

Differentially private (DP) image synthesis aims to generate synthetic images from a sensitive dataset, alleviating the privacy leakage concerns of organizations sharing and utilizing synthetic images. Although previous methods have significantly progressed, especially in training diffusion models on sensitive images with DP Stochastic Gradient Descent (DP-SGD), they still suffer from unsatisfactory performance. In this work, inspired by curriculum learning, we propose a two-stage DP image synthesis framework, where diffusion models learn to generate DP synthetic images from easy to hard. Unlike existing methods that directly use DP-SGD to train diffusion models, we propose an easy stage in the beginning, where diffusion models learn simple features of the sensitive images. To facilitate this easy stage, we propose to use `central images', simply aggregations of random samples of the sensitive dataset. Intuitively, although those central images do not show details, they demonstrate useful characteristics of all images and only incur minimal privacy costs, thus helping early-phase model training. We conduct experiments to present that on the average of four investigated image datasets, the fidelity and utility metrics of our synthetic images are 33.1% and 2.1% better than the state-of-the-art method.

cs.CR↗

arXiv2025

Tianyi: A Traditional Chinese Medicine all-rounder language model and its Real-World Clinical Practice

Natural medicines, particularly Traditional Chinese Medicine (TCM), are gaining global recognition for their therapeutic potential in addressing human symptoms and diseases. TCM, with its systematic theories and extensive practical experience, provides abundant resources for healthcare. However, the effective application of TCM requires precise syndrome diagnosis, determination of treatment principles, and prescription formulation, which demand decades of clinical expertise. Despite advancements in TCM-based decision systems, machine learning, and deep learning research, limitations in data and single-objective constraints hinder their practical application. In recent years, large language models (LLMs) have demonstrated potential in complex tasks, but lack specialization in TCM and face significant challenges, such as too big model scale to deploy and issues with hallucination. To address these challenges, we introduce Tianyi with 7.6-billion-parameter LLM, a model scale proper and specifically designed for TCM, pre-trained and fine-tuned on diverse TCM corpora, including classical texts, expert treatises, clinical records, and knowledge graphs. Tianyi is designed to assimilate interconnected and systematic TCM knowledge through a progressive learning manner. Additionally, we establish TCMEval, a comprehensive evaluation benchmark, to assess LLMs in TCM examinations, clinical tasks, domain-specific question-answering, and real-world trials. The extensive evaluations demonstrate the significant potential of Tianyi as an AI assistant in TCM clinical practice and research, bridging the gap between TCM knowledge and practical application.

cs.CL↗

arXiv2025

LREA: Low-Rank Efficient Attention on Modeling Long-Term User Behaviors for CTR Prediction

With the rapid growth of user historical behavior data, user interest modeling has become a prominent aspect in Click-Through Rate (CTR) prediction, focusing on learning user intent representations. However, this complexity poses computational challenges, requiring a balance between model performance and acceptable response times for online services. Traditional methods often utilize filtering techniques. These techniques can lead to the loss of significant information by prioritizing top K items based on item attributes or employing low-precision attention mechanisms. In this study, we introduce LREA, a novel attention mechanism that overcomes the limitations of existing approaches while ensuring computational efficiency. LREA leverages low-rank matrix decomposition to optimize runtime performance and incorporates a specially designed loss function to maintain attention capabilities while preserving information integrity. During the inference phase, matrix absorption and pre-storage strategies are employed to effectively meet runtime constraints. The results of extensive offline and online experiments demonstrate that our method outperforms state-of-the-art approaches.

cs.IR↗

arXiv2025

Tracing Content Requirements in Financial Documents using Multi-granularity Text Analysis

The completeness (in terms of content) of financial documents is a fundamental requirement for investment funds. To ensure completeness, financial regulators have to spend significant time carefully checking every financial document based on relevant content requirements, which prescribe the information types to be included in financial documents (e.g., the description of shares' issue conditions and procedures). However, existing techniques provide limited support to help regulators automatically identify the text chunks related to financial information types, due to the complexity of financial documents. In this paper, we propose FITI to trace content requirements in financial documents with multi-granularity text analysis. Given a new financial document, FITI first selects a set of candidate sentences for efficient information type identification. Then, to rank candidate sentences, FITI uses a combination of rule-based and data-centric approaches, by leveraging information retrieval (IR) and machine learning (ML) techniques that analyze the words, sentences, and contexts related to an information type. Finally, a heuristic-based selector, which considers both the sentence ranking and domain-specific phrases, determines a list of sentences corresponding to each information type. We evaluated FITI by assessing its effectiveness in tracing financial content requirements in 100 real-world financial documents. Experimental results show that FITI is able to provide accurate identification with average precision, recall, and F1-score values of 0.824, 0.646, and 0.716, respectively. The overall accuracy of FITI significantly outperforms the best baseline (based on a transformer language model) by 0.266 in terms of F1-score. Furthermore, FITI can help regulators detect about 80% of missing information types in financial documents

cs.IR↗

arXiv2025

Addressing Information Loss and Interaction Collapse: A Dual Enhanced Attention Framework for Feature Interaction

The Transformer has proven to be a significant approach in feature interaction for CTR prediction, achieving considerable success in previous works. However, it also presents potential challenges in handling feature interactions. Firstly, Transformers may encounter information loss when capturing feature interactions. By relying on inner products to represent pairwise relationships, they compress raw interaction information, which can result in a degradation of fidelity. Secondly, due to the long-tail features distribution, feature fields with low information-abundance embeddings constrain the information abundance of other fields, leading to collapsed embedding matrices. To tackle these issues, we propose a Dual Attention Framework for Enhanced Feature Interaction, known as Dual Enhanced Attention. This framework integrates two attention mechanisms: the Combo-ID attention mechanism and the collapse-avoiding attention mechanism. The Combo-ID attention mechanism directly retains feature interaction pairs to mitigate information loss, while the collapse-avoiding attention mechanism adaptively filters out low information-abundance interaction pairs to prevent interaction collapse. Extensive experiments conducted on industrial datasets have shown the effectiveness of Dual Enhanced Attention.

cs.IR↗

arXiv2025

A Novel Interactive-Guided Differential Testing Approach for FPGA Simulation Debugger Tools

Field-Programmable Gate Array (FPGA) development tool chains are widely used in FPGA design, simulation, and verification in critical areas like communications, automotive electronics, and aerospace. Commercial FPGA tool chains such as Xilinx' Vivado aids developers in swiftly identifying and rectifying bugs and issues in FPGA designs through a robust built-in debugger, ensuring the correctness and development efficiency of the FPGA design. Hardening such FPGA chip debugger tools by testing is crucial since engineers might misinterpret code and introduce incorrect fixes, leading to security risks. However, FPGA chip debugger tools are challenging to test as they require assessing both RTL designs and a series of debugging actions, including setting breakpoints and stepping through the code. To address this issue, we propose a interactive differential testing approach called DB-Hunter to detect bugs in Vivado's FPGA chip debugger tools. Specifically, DB-Hunter consists of three components: RTL design transformation component, debug action transformation component, and interactive differential testing component. By performing RTL design and debug action transformations, DB-Hunter generates diverse and complex RTL designs and debug actions, to thoroughly test the Vivado debugger using interactive differential testing to detect bugs. In three months, DB-Hunter reported 18 issues, including 10 confirmed as bugs by Xilinx Support, 6 bugs had been fixed in last version.

cs.SE↗

arXiv2025

Planetarium: A Rigorous Benchmark for Translating Text to Structured Planning Languages

Recent works have explored using language models for planning problems. One approach examines translating natural language descriptions of planning tasks into structured planning languages, such as the planning domain definition language (PDDL). Existing evaluation methods struggle to ensure semantic correctness and rely on simple or unrealistic datasets. To bridge this gap, we introduce \textit{Planetarium}, a benchmark designed to evaluate language models' ability to generate PDDL code from natural language descriptions of planning tasks. \textit{Planetarium} features a novel PDDL equivalence algorithm that flexibly evaluates the correctness of generated PDDL, along with a dataset of 145,918 text-to-PDDL pairs across 73 unique state combinations with varying levels of difficulty. Finally, we evaluate several API-access and open-weight language models that reveal this task's complexity. For example, 96.1\% of the PDDL problem descriptions generated by GPT-4o are syntactically parseable, 94.4\% are solvable, but only 24.8\% are semantically correct, highlighting the need for a more rigorous benchmark for this problem.

cs.CL↗

arXiv2024

Preference Tuning For Toxicity Mitigation Generalizes Across Languages

Detoxifying multilingual Large Language Models (LLMs) has become crucial due to their increasing global use. In this work, we explore zero-shot cross-lingual generalization of preference tuning in detoxifying LLMs. Unlike previous studies that show limited cross-lingual generalization for other safety tasks, we demonstrate that Direct Preference Optimization (DPO) training with only English data can significantly reduce toxicity in multilingual open-ended generations. For example, the probability of mGPT-1.3B generating toxic continuations drops from 46.8% to 3.9% across 17 different languages after training. Our results also extend to other multilingual LLMs, such as BLOOM, Llama3, and Aya-23. Using mechanistic interpretability tools like causal intervention and activation analysis, we identified the dual multilinguality property of MLP layers in LLMs, which explains the cross-lingual generalization of DPO. Finally, we show that bilingual sentence retrieval can predict the cross-lingual transferability of DPO preference tuning.

cs.CL↗

arXiv2024

AdaFlow: Opportunistic Inference on Asynchronous Mobile Data with Generalized Affinity Control

The rise of mobile devices equipped with numerous sensors, such as LiDAR and cameras, has spurred the adoption of multi-modal deep intelligence for distributed sensing tasks, such as smart cabins and driving assistance. However, the arrival times of mobile sensory data vary due to modality size and network dynamics, which can lead to delays (if waiting for slower data) or accuracy decline (if inference proceeds without waiting). Moreover, the diversity and dynamic nature of mobile systems exacerbate this challenge. In response, we present a shift to \textit{opportunistic} inference for asynchronous distributed multi-modal data, enabling inference as soon as partial data arrives. While existing methods focus on optimizing modality consistency and complementarity, known as modal affinity, they lack a \textit{computational} approach to control this affinity in open-world mobile environments. AdaFlow pioneers the formulation of structured cross-modality affinity in mobile contexts using a hierarchical analysis-based normalized matrix. This approach accommodates the diversity and dynamics of modalities, generalizing across different types and numbers of inputs. Employing an affinity attention-based conditional GAN (ACGAN), AdaFlow facilitates flexible data imputation, adapting to various modalities and downstream tasks without retraining. Experiments show that AdaFlow significantly reduces inference latency by up to 79.9\% and enhances accuracy by up to 61.9\%, outperforming status quo approaches.

cs.LG↗

arXiv2024

Deep Learning-based Software Engineering: Progress, Challenges, and Opportunities

Researchers have recently achieved significant advances in deep learning techniques, which in turn has substantially advanced other research disciplines, such as natural language processing, image processing, speech recognition, and software engineering. Various deep learning techniques have been successfully employed to facilitate software engineering tasks, including code generation, software refactoring, and fault localization. Many papers have also been presented in top conferences and journals, demonstrating the applications of deep learning techniques in resolving various software engineering tasks. However, although several surveys have provided overall pictures of the application of deep learning techniques in software engineering, they focus more on learning techniques, that is, what kind of deep learning techniques are employed and how deep models are trained or fine-tuned for software engineering tasks. We still lack surveys explaining the advances of subareas in software engineering driven by deep learning techniques, as well as challenges and opportunities in each subarea. To this end, in this paper, we present the first task-oriented survey on deep learning-based software engineering. It covers twelve major software engineering subareas significantly impacted by deep learning techniques. Such subareas spread out the through the whole lifecycle of software development and maintenance, including requirements engineering, software development, testing, maintenance, and developer collaboration. As we believe that deep learning may provide an opportunity to revolutionize the whole discipline of software engineering, providing one survey covering as many subareas as possible in software engineering can help future research push forward the frontier of deep learning-based software engineering more systematically.

cs.SE↗

arXiv2024

DPI: Ensuring Strict Differential Privacy for Infinite Data Streaming

Streaming data, crucial for applications like crowdsourcing analytics, behavior studies, and real-time monitoring, faces significant privacy risks due to the large and diverse data linked to individuals. In particular, recent efforts to release data streams, using the rigorous privacy notion of differential privacy (DP), have encountered issues with unbounded privacy leakage. This challenge limits their applicability to only a finite number of time slots (''finite data stream'') or relaxation to protecting the events (''event or $w$-event DP'') rather than all the records of users. A persistent challenge is managing the sensitivity of outputs to inputs in situations where users contribute many activities and data distributions evolve over time. In this paper, we present a novel technique for Differentially Private data streaming over Infinite disclosure (DPI) that effectively bounds the total privacy leakage of each user in infinite data streams while enabling accurate data collection and analysis. Furthermore, we also maximize the accuracy of DPI via a novel boosting mechanism. Finally, extensive experiments across various streaming applications and real datasets (e.g., COVID-19, Network Traffic, and USDA Production), show that DPI maintains high utility for infinite data streams in diverse settings. Code for DPI is available at https://github.com/ShuyaFeng/DPI.

cs.CR↗

arXiv2024

A Novel HDL Code Generator for Effectively Testing FPGA Logic Synthesis Compilers

Field Programmable Gate Array (FPGA) logic synthesis compilers (e.g., Vivado, Iverilog, Yosys, and Quartus) are widely applied in Electronic Design Automation (EDA), such as the development of FPGA programs.However, defects (i.e., incorrect synthesis) in logic synthesis compilers may lead to unexpected behaviors in target applications, posing security risks. Therefore, it is crucial to thoroughly test logic synthesis compilers to eliminate such defects.Despite several Hardware Design Language (HDL) code generators (e.g., Verismith) have been proposed to find defects in logic synthesis compilers, the effectiveness of these generators is still limited by the simple code generation strategy and the monogeneity of the generated HDL code.This paper proposes LegoHDL, a novel method to generate syntax valid HDL code for comprehensively testing FPGA logic synthesis compilers.LegoHDL can generate more complex and diverse defect-trigger HDL code (e.g., Verilog, VHDL, and SystemVerilog) by leveraging the guidance of abstract syntax tree and the extensive function block libraries of cyber-physical systems. Extensive experiments show that the diversity and defect-trigger capability of HDL code generated by LegoHDL are significantly better than the state-of-the-art method (i.e., Verismith).In three months, LegoHDL has reported 20 new defects--many of which are deep and important; 16 of them have been confirmed.

cs.AR↗

arXiv2024

EchoPFL: Asynchronous Personalized Federated Learning on Mobile Devices with On-Demand Staleness Control

The rise of mobile devices with abundant sensory data and local computing capabilities has driven the trend of federated learning (FL) on these devices. And personalized FL (PFL) emerges to train specific deep models for each mobile device to address data heterogeneity and varying performance preferences. However, mobile training times vary significantly, resulting in either delay (when waiting for slower devices for aggregation) or accuracy decline (when aggregation proceeds without waiting). In response, we propose a shift towards asynchronous PFL, where the server aggregates updates as soon as they are available. Nevertheless, existing asynchronous protocols are unfit for PFL because they are devised for federated training of a single global model. They suffer from slow convergence and decreased accuracy when confronted with severe data heterogeneity prevalent in PFL. Furthermore, they often exclude slower devices for staleness control, which notably compromises accuracy when these devices possess critical personalized data. Therefore, we propose EchoPFL, a coordination mechanism for asynchronous PFL. Central to EchoPFL is to include updates from all mobile devices regardless of their latency. To cope with the inevitable staleness from slow devices, EchoPFL revisits model broadcasting. It intelligently converts the unscalable broadcast to on-demand broadcast, leveraging the asymmetrical bandwidth in wireless networks and the dynamic clustering-based PFL. Experiments show that compared to status quo approaches, EchoPFL achieves a reduction of up to 88.2% in convergence time, an improvement of up to 46% in accuracy, and a decrease of 37% in communication costs

cs.DC↗

arXiv2023

Learning-Based Relaxation of Completeness Requirements for Data Entry Forms

Data entry forms use completeness requirements to specify the fields that are required or optional to fill for collecting necessary information from different types of users. However, some required fields may not be applicable for certain types of users anymore. Nevertheless, they may still be incorrectly marked as required in the form; we call such fields obsolete required fields. Since obsolete required fields usually have not-null validation checks before submitting the form, users have to enter meaningless values in such fields in order to complete the form submission. These meaningless values threaten the quality of the filled data. To avoid users filling meaningless values, existing techniques usually rely on manually written rules to identify the obsolete required fields and relax their completeness requirements. However, these techniques are ineffective and costly. In this paper, we propose LACQUER, a learning-based automated approach for relaxing the completeness requirements of data entry forms. LACQUER builds Bayesian Network models to automatically learn conditions under which users had to fill meaningless values. To improve its learning ability, LACQUER identifies the cases where a required field is only applicable for a small group of users, and uses SMOTE, an oversampling technique, to generate more instances on such fields for effectively mining dependencies on them. Our experimental results show that LACQUER can accurately relax the completeness requirements of required fields in data entry forms with precision values ranging between 0.76 and 0.90 on different datasets. LACQUER can prevent users from filling 20% to 64% of meaningless values, with negative predictive values between 0.72 and 0.91. Furthermore, LACQUER is efficient; it takes at most 839 ms to predict the completeness requirement of an instance.

cs.SE↗

arXiv2023

Local Differentially Private Heavy Hitter Detection in Data Streams with Bounded Memory

Top-$k$ frequent items detection is a fundamental task in data stream mining. Many promising solutions are proposed to improve memory efficiency while still maintaining high accuracy for detecting the Top-$k$ items. Despite the memory efficiency concern, the users could suffer from privacy loss if participating in the task without proper protection, since their contributed local data streams may continually leak sensitive individual information. However, most existing works solely focus on addressing either the memory-efficiency problem or the privacy concerns but seldom jointly, which cannot achieve a satisfactory tradeoff between memory efficiency, privacy protection, and detection accuracy. In this paper, we present a novel framework HG-LDP to achieve accurate Top-$k$ item detection at bounded memory expense, while providing rigorous local differential privacy (LDP) protection. Specifically, we identify two key challenges naturally arising in the task, which reveal that directly applying existing LDP techniques will lead to an inferior ``accuracy-privacy-memory efficiency'' tradeoff. Therefore, we instantiate three advanced schemes under the framework by designing novel LDP randomization methods, which address the hurdles caused by the large size of the item domain and by the limited space of the memory. We conduct comprehensive experiments on both synthetic and real-world datasets to show that the proposed advanced schemes achieve a superior ``accuracy-privacy-memory efficiency'' tradeoff, saving $2300\times$ memory over baseline methods when the item domain size is $41,270$. Our code is open-sourced via the link.

cs.CR↗

arXiv2023

Abstract-to-Executable Trajectory Translation for One-Shot Task Generalization

Training long-horizon robotic policies in complex physical environments is essential for many applications, such as robotic manipulation. However, learning a policy that can generalize to unseen tasks is challenging. In this work, we propose to achieve one-shot task generalization by decoupling plan generation and plan execution. Specifically, our method solves complex long-horizon tasks in three steps: build a paired abstract environment by simplifying geometry and physics, generate abstract trajectories, and solve the original task by an abstract-to-executable trajectory translator. In the abstract environment, complex dynamics such as physical manipulation are removed, making abstract trajectories easier to generate. However, this introduces a large domain gap between abstract trajectories and the actual executed trajectories as abstract trajectories lack low-level details and are not aligned frame-to-frame with the executed trajectory. In a manner reminiscent of language translation, our approach leverages a seq-to-seq model to overcome the large domain gap between the abstract and executable trajectories, enabling the low-level policy to follow the abstract trajectory. Experimental results on various unseen long-horizon tasks with different robot embodiments demonstrate the practicability of our methods to achieve one-shot task generalization.

cs.LG↗