arXiv subjects

Siyi Wang

Publications and source records attributed to Siyi Wang.

At least 37 records · Page 2Linked to original sources

arXiv2026

Scaling Ambiguity: Augmenting Human Annotation in Speech Emotion Recognition with Audio-Language Models

Speech Emotion Recognition models typically use single categorical labels, overlooking the inherent ambiguity of human emotions. Ambiguous Emotion Recognition addresses this by representing emotions as probability distributions, but progress is limited by unreliable ground-truth distributions inferred from sparse human annotations. This paper explores whether Large Audio-Language Models (ALMs) can mitigate the annotation bottleneck by generating high-quality synthetic annotations. We introduce a framework leveraging ALMs to create Synthetic Perceptual Proxies, augmenting human annotations to improve ground-truth distribution reliability. We validate these proxies through statistical analysis of their alignment with human distributions and evaluate their impact by fine-tuning ALMs with the augmented emotion distributions. Furthermore, to address class imbalance and enable unbiased evaluation, we propose DiME-Aug, a Distribution-aware Multimodal Emotion Augmentation strategy. Experiments on IEMOCAP and MSP-Podcast show that synthetic annotations enhance emotion distribution, especially in low-ambiguity regions where annotation agreement is high. However, benefits diminish for highly ambiguous emotions with greater human disagreement. This work provides the first evidence that ALMs could address annotation scarcity in ambiguous emotion recognition, but highlights the need for more advanced prompting or generation strategies to handle highly ambiguous cases.

eess.AS↗

arXiv2025

Risk-Averse Learning with Varying Risk Levels

In safety-critical decision-making, the environment may evolve over time, and the learner adjusts its risk level accordingly. This work investigates risk-averse online optimization in dynamic environments with varying risk levels, employing Conditional Value-at-Risk (CVaR) as the risk measure. To capture the dynamics of the environment and risk levels, we employ the function variation metric and introduce a novel risk-level variation metric. Two information settings are considered: a first-order scenario, where the learner observes both function values and their gradients; and a zeroth-order scenario, where only function evaluations are available. For both cases, we develop risk-averse learning algorithms with a limited sampling budget and analyze their dynamic regret bounds in terms of function variation, risk-level variation, and the total number of samples. The regret analysis demonstrates the adaptability of the algorithms in non-stationary and risk-sensitive settings. Finally, numerical experiments are presented to demonstrate the efficacy of the methods.

math.OC↗

arXiv2025

Preference-based optimization from noisy pairwise comparisons

In interactive systems, feedback is often provided in the form of preference between queried options rather than precise scores, which motivates optimization methods to learn from such comparisons. In this work, we propose a preference-based optimization algorithm that relies on noisy two-point comparisons. At each iteration, the algorithm employs a uniform-sphere perturbation to generate a perturbed action and queries the resulting loss comparison to estimate a descent direction. We demonstrate that, under standard smoothness and bounded variance assumptions, the algorithm converges to a stationary point when the smoothing and step size parameters are properly chosen. Numerical experiments on an LQG system demonstrate the effectiveness of the preference-based optimization algorithm with comparison feedback.

math.OC↗

arXiv2025

Decompile-Bench: Million-Scale Binary-Source Function Pairs for Real-World Binary Decompilation

Recent advances in LLM-based decompilers have been shown effective to convert low-level binaries into human-readable source code. However, there still lacks a comprehensive benchmark that provides large-scale binary-source function pairs, which is critical for advancing the LLM decompilation technology. Creating accurate binary-source mappings incurs severe issues caused by complex compilation settings and widespread function inlining that obscure the correspondence between binaries and their original source code. Previous efforts have either relied on used contest-style benchmarks, synthetic binary-source mappings that diverge significantly from the mappings in real world, or partially matched binaries with only code lines or variable names, compromising the effectiveness of analyzing the binary functionality. To alleviate these issues, we introduce Decompile-Bench, the first open-source dataset comprising two million binary-source function pairs condensed from 100 million collected function pairs, i.e., 450GB of binaries compiled from permissively licensed GitHub projects. For the evaluation purposes, we also developed a benchmark Decompile-Bench-Eval including manually crafted binaries from the well-established HumanEval and MBPP, alongside the compiled GitHub repositories released after 2025 to mitigate data leakage issues. We further explore commonly-used evaluation metrics to provide a thorough assessment of the studied LLM decompilers and find that fine-tuning with Decompile-Bench causes a 20% improvement over previous benchmarks in terms of the re-executability rate. Our code and data has been released in HuggingFace and Github. https://github.com/albertan017/LLM4Decompile

cs.SE↗

arXiv2025

SK2Decompile: LLM-based Two-Phase Binary Decompilation from Skeleton to Skin

Large Language Models (LLMs) have emerged as a promising approach for binary decompilation. However, the existing LLM-based decompilers still are somewhat limited in effectively presenting a program's source-level structure with its original identifiers. To mitigate this, we introduce SK2Decompile, a novel two-phase approach to decompile from the skeleton (semantic structure) to the skin (identifier) of programs. Specifically, we first apply a Structure Recovery model to translate a program's binary code to an Intermediate Representation (IR) as deriving the program's "skeleton", i.e., preserving control flow and data structures while obfuscating all identifiers with generic placeholders. We also apply reinforcement learning to reward the model for producing program structures that adhere to the syntactic and semantic rules expected by compilers. Second, we apply an Identifier Naming model to produce meaningful identifiers which reflect actual program semantics as deriving the program's "skin". We train the Identifier Naming model with a separate reinforcement learning objective that rewards the semantic similarity between its predictions and the reference code. Such a two-phase decompilation process facilitates advancing the correctness and readability of decompilation independently. Our evaluations indicate that SK2Decompile, significantly outperforms the SOTA baselines, achieving 21.6% average re-executability rate gain over GPT-5-mini on the HumanEval dataset and 29.4% average R2I improvement over Idioms on the GitHub2025 benchmark.

cs.SE↗

arXiv2025

Risk-averse learning with delayed feedback

In real-world scenarios, risk-averse learning is valuable for mitigating potential adverse outcomes. However, the delayed feedback makes it challenging to assess and manage risk effectively. In this paper, we investigate risk-averse learning using Conditional Value at Risk (CVaR) as risk measure, while incorporating feedback with random but bounded delays. We develop two risk-averse learning algorithms that rely on one-point and two-point zeroth-order optimization approaches, respectively. The dynamic regrets of the algorithms are analyzed in terms of the cumulative delay and the number of total samplings. In the absence of delay, the regret bounds match the established bounds of zeroth-order stochastic gradient methods for risk-averse learning. Furthermore, the two-point risk-averse learning outperforms the one-point algorithm by achieving a smaller regret bound. We provide numerical experiments on a dynamic pricing problem to demonstrate the performance of the algorithms.

cs.LG↗

arXiv2025

An Investigation into Value Misalignment in LLM-Generated Texts for Cultural Heritage

As Large Language Models (LLMs) become increasingly prevalent in tasks related to cultural heritage, such as generating descriptions of historical monuments, translating ancient texts, preserving oral traditions, and creating educational content, their ability to produce accurate and culturally aligned texts is being increasingly relied upon by users and researchers. However, cultural value misalignments may exist in generated texts, such as the misrepresentation of historical facts, the erosion of cultural identity, and the oversimplification of complex cultural narratives, which may lead to severe consequences. Therefore, investigating value misalignment in the context of LLM for cultural heritage is crucial for mitigating these risks, yet there has been a significant lack of systematic and comprehensive study and investigation in this area. To fill this gap, we systematically assess the reliability of LLMs in generating culturally aligned texts for cultural heritage-related tasks. We conduct a comprehensive evaluation by compiling an extensive set of 1066 query tasks covering 5 widely recognized categories with 17 aspects within the knowledge framework of cultural heritage across 5 open-source LLMs, and examine both the type and rate of cultural value misalignments in the generated texts. Using both automated and manual approaches, we effectively detect and analyze the cultural value misalignments in LLM-generated texts. Our findings are concerning: over 65% of the generated texts exhibit notable cultural misalignments, with certain tasks demonstrating almost complete misalignment with key cultural values. Beyond these findings, this paper introduces a benchmark dataset and a comprehensive evaluation workflow that can serve as a valuable resource for future research aimed at enhancing the cultural sensitivity and reliability of LLMs.

cs.CL↗

arXiv2025

$β$-integrated local depth and corresponding partitioned local depth representation

A novel local depth definition, $β$-integrated local depth ($β$-ILD), is proposed as a generalization of the local depth introduced by Paindaveine and Van Bever \cite{paindaveine2013depth}, designed to quantify the local centrality of data points. $β$-ILD inherits desirable properties from global data depth and remains robust across varying locality levels. A partitioning approach for $β$-ILD is introduced, leading to the construction of a matrix that quantifies the contribution of one point to another's local depth, providing a new interpretable measure of local centrality. These concepts are applied to classification and outlier detection tasks, demonstrating significant improvements in the performance of depth-based algorithms.

math.ST↗

arXiv2025

Token-Level Logits Matter: A Closer Look at Speech Foundation Models for Ambiguous Emotion Recognition

Emotional intelligence in conversational AI is crucial across domains like human-computer interaction. While numerous models have been developed, they often overlook the complexity and ambiguity inherent in human emotions. In the era of large speech foundation models (SFMs), understanding their capability in recognizing ambiguous emotions is essential for the development of next-generation emotion-aware models. This study examines the effectiveness of SFMs in ambiguous emotion recognition. We designed prompts for ambiguous emotion prediction and introduced two novel approaches to infer ambiguous emotion distributions: one analysing generated text responses and the other examining the internal processing of SFMs through token-level logits. Our findings suggest that while SFMs may not consistently generate accurate text responses for ambiguous emotions, they can interpret such emotions at the token level based on prior knowledge, demonstrating robustness across different prompts.

cs.SD↗

arXiv2025

SWAP Attack: Stealthy Side-Channel Attack on Multi-Tenant Quantum Cloud System

The rapid advancement of quantum computing has spurred widespread adoption, with cloud-based quantum devices gaining traction in academia and industry. This shift raises critical concerns about the privacy and security of computations on shared, multi-tenant quantum platforms accessed remotely. Recent studies have shown that crosstalk on shared quantum devices allows adversaries to interfere with victim circuits within a neighborhood. While insightful, these works left unresolved questions regarding the root cause of crosstalk, effective countermeasures, and replicability across circuits. We revisit the crosstalk effect, tracing its origins to the SWAP path between qubits and demonstrating its impact even over long distances. Our results significantly improve the understanding of this phenomenon beyond prior works. The proposed SWAP-based side-channel attack operates in both active and passive modes, as verified on real IBM quantum devices. In the active attack, an attacker executing a single CNOT gate can perturb victim circuits running Grover's Algorithm, reducing expected output accuracy by $81.62\%$ through strategic qubit placement. Moreover, this effect can be modeled to identify qubits more susceptible to attack. The passive attack, leveraging a stealthy circuit as small as $6.25\%$ of the victim's, achieves $100\%$ accuracy in predicting the victim's circuit size when running Simon's Algorithm. These findings challenge the existing defense strategy of maximizing topological distance between circuits, showing that attackers can still extract sensitive information or manipulate results remotely. Our work highlights the urgent need for robust security measures to safeguard quantum computations against emerging threats.

quant-ph↗

arXiv2025

On Exact Space-Depth Trade-Offs in Multi-Controlled Toffoli Decomposition

In this paper, we consider the optimized implementation of Multi Controlled Toffoli (MCT) using the Clifford $+$ T gate sets. While there are several recent works in this direction, here we explicitly quantify the trade-off (with concrete formulae) between the Toffoli depth (this means the depth using the classical 2-controlled Toffoli) of the $n$-controlled Toffoli (hereform we will tell $n$-MCT) and the number of clean ancilla qubits. Additionally, we achieve a reduced Toffoli depth (and consequently, T-depth), which is an extension of the technique introduced by Khattar et al. (2024). In terms of a negative result, we first show that using such conditionally clean ancilla techniques, Toffoli depth can never achieve exactly $\ceil{\log_2 n}$, though it remains of the same order. This highlights the limitation of the techniques exploiting conditionally clean ancilla [Nie et al., 2024, Khattar et al., 2024]. Then we prove that, in a more general setup, the T-Depth in the Clifford + T decomposition, via Toffoli gates, is lower bounded by $\ceil{\log_2 n}$, and this bound is achieved following the complete binary tree structure. Since the ($2$-controlled) Toffoli gate can further be decomposed using Clifford $+$ T, various methodologies are explored too in this regard for trade-off related implications.

quant-ph↗

arXiv2025

Infinite-horizon optimal scheduling for feedback control

Emerging cyber-physical systems impel the development of communication protocols that optimize resource utilization. This article investigates infinite-horizon optimal scheduling for resource-aware networked control systems by addressing the rate-regulation tradeoff. Consider a scenario where the sensor and the controller communicate via a networked channel, the transmission scheduling problem is formulated as a Markov decision process on unbounded general state space controlled by scheduling decisions. The value of information (VoI) serves as a metric to assess the importance of sensory data for transmission. We derive the optimal scheduling law for feedback control based on VoI and show that it is deterministic and stationary, with an explicit expression obtained via value iteration. The closed-loop system under the designed scheduling law is shown to be stochastically stable. By analyzing the dynamic behavior of the iteration process, we show that the VoI function and the optimal scheduling law exhibit symmetry. Furthermore, when the system matrix is diagonalizable, the VoI function is monotone and quasi-convex. Consequently, the optimal scheduling law is shown to exhibit a threshold structure and takes a quadratic form, with the threshold region explicitly characterized. Finally, the numerical simulation illustrates the theoretical result of the VoI-based scheduling.

eess.SY↗

arXiv2025

Optimal Transmission Power Scheduling for Networked Control System under DoS Attack

Designing networked control systems that are reliable and resilient against adversarial threats, is essential for ensuring the security of cyber-physical systems. This paper addresses the communication-control co-design problem for networked control systems under denial-of-service (DoS) attacks. In the wireless channel, a transmission power scheduler periodically determines the power level for sensory data transmission. Yet DoS attacks render data packets unavailable by disrupting the communication channel. This paper co-designs the control and power scheduling laws in the presence of DoS attacks and aims to minimize the sum of regulation control performance and transmission power consumption. Both finite- and infinite-horizon discounted cost criteria are addressed, respectively. By delving into the information structure between the controller and the power scheduler under attack, the original co-design problem is divided into two subproblems that can be solved individually without compromising optimality. The optimal control is shown to be certainty equivalent, and the optimal transmission power scheduling is solved using a dynamic programming approach. Moreover, in the infinite-horizon scenario, we analyze the performance of the designed scheduling policy and develop an upper bound of the total costs. Finally, a numerical example is provided to demonstrate the theoretical results.

eess.SY↗

arXiv2024

Hybrid Cooperative Co-Evolution Algorithm for Deadlock-prone Distributed Assembly Flowshop Scheduling with Limited buffers Using Petri nets

The distributed assembly flowshop scheduling problem (DAFSP) can be applied to immense manufacturing environments. In DAFSP, jobs are first processed in distributed flowshops, and then assembled into final products by an assembly machine, which usually has limited buffers in practical application. This limited capacity can lead to deadlocks, halting job completion and blocking the entire manufacturing process. However, existing scheduling methods fail to address these deadlocks in DAFSP effectively. As such, we develop a hybrid cooperative co-evolution (HCCE) algorithm for solving the deadlock-prone DAFSP by minimizing the makespan. For the first time, we use Petri nets to analyze the deadlocks in DAFSP and propose a Petri net-based deadlock amending method (IDAM), which is further integrated into HCCE to ensure the feasibility (i.e., deadlock-freeness) of solutions. Importantly, HCCE contains an elite archive (EAR) and two subpopulations. It uses the problem-specific operators for heuristic initialization and global-search. To enhance the quality and diversity of solutions, an information transfer mechanism (ITM) is developed among subpopulation and EAR, and four local-search operators are performed sequentially on each individual in EAR. Finally, comprehensive experiments demonstrate the effectiveness and superiority of the proposed HCCE algorithm.

cs.DC↗

arXiv2024

Unveiling the Potential of Robustness in Selecting Conditional Average Treatment Effect Estimators

The growing demand for personalized decision-making has led to a surge of interest in estimating the Conditional Average Treatment Effect (CATE). Various types of CATE estimators have been developed with advancements in machine learning and causal inference. However, selecting the desirable CATE estimator through a conventional model validation procedure remains impractical due to the absence of counterfactual outcomes in observational data. Existing approaches for CATE estimator selection, such as plug-in and pseudo-outcome metrics, face two challenges. First, they must determine the metric form and the underlying machine learning models for fitting nuisance parameters (e.g., outcome function, propensity function, and plug-in learner). Second, they lack a specific focus on selecting a robust CATE estimator. To address these challenges, this paper introduces a Distributionally Robust Metric (DRM) for CATE estimator selection. The proposed DRM is nuisance-free, eliminating the need to fit models for nuisance parameters, and it effectively prioritizes the selection of a distributionally robust CATE estimator. The experimental results validate the effectiveness of the DRM method in selecting CATE estimators that are robust to the distribution shift incurred by covariate shift and hidden confounders.

cs.LG↗

arXiv2024

Leveraging Large Vision Language Model For Better Automatic Web GUI Testing

With the rapid development of web technology, more and more software applications have become web-based in the past decades. To ensure software quality and user experience, various techniques have been proposed to automatically test web applications by interacting with their GUIs. To achieve high functional coverage, web GUI testing tools often need to generate high-quality text inputs and interact with the associated GUI elements (e.g., click submit buttons). However, developing a holistic approach that solves both subtasks is challenging because the web GUI context can be complicated and highly dynamic, which makes it hard to process programmatically. The recent development of large vision-language models (LVLM) provides new opportunities to handle these longstanding problems. This paper proposes VETL, the first LVLM-driven end-to-end web testing technique. With LVLM's scene understanding capabilities, VETL can generate valid and meaningful text inputs focusing on the local context, while avoiding the need to extract precise textual attributes. The selection of associated GUI elements is formulated as a visual question-answering problem, allowing LVLM to capture the logical connection between the input box and the relevant element based on visual instructions. Further, the GUI exploration is guided by a multi-armed bandit module employing a curiosity-oriented strategy. Experiments show that VETL effectively explores web state/action spaces and detects bugs. Compared with WebExplor, the state-of-the-art web testing technique, VETL can discover 25% more unique web actions on benchmark websites. Moreover, it can expose functional bugs in top-ranking commercial websites, which the website maintainers have confirmed. Our work makes the first attempt at leveraging LVLM in end-to-end GUI testing, demonstrating promising results in this research direction.

cs.SE↗

arXiv2024

Diffusion-based Speech Enhancement with Schrödinger Bridge and Symmetric Noise Schedule

Recently, diffusion-based generative models have demonstrated remarkable performance in speech enhancement tasks. However, these methods still encounter challenges, including the lack of structural information and poor performance in low Signal-to-Noise Ratio (SNR) scenarios. To overcome these challenges, we propose the Schröodinger Bridge-based Speech Enhancement (SBSE) method, which learns the diffusion processes directly between the noisy input and the clean distribution, unlike conventional diffusion-based speech enhancement systems that learn data to Gaussian distributions. To enhance performance in extremely noisy conditions, we introduce a two-stage system incorporating ratio mask information into the diffusion-based generative model. Our experimental results show that our proposed SBSE method outperforms all the baseline models and achieves state-of-the-art performance, especially in low SNR conditions. Importantly, only a few inference steps are required to achieve the best result.

eess.AS↗

arXiv2024

A Comprehensive Study of Quantum Arithmetic Circuits

In recent decades, the field of quantum computing has experienced remarkable progress. This progress is marked by the superior performance of many quantum algorithms compared to their classical counterparts, with Shor's algorithm serving as a prominent illustration. Quantum arithmetic circuits, which are the fundamental building blocks in numerous quantum algorithms, have attracted much attention. Despite extensive exploration of various designs in the existing literature, researchers remain keen on developing novel designs and improving existing ones. In this review article, we aim to provide a systematically organized and easily comprehensible overview of the current state-of-the-art in quantum arithmetic circuits. Specifically, this study covers fundamental operations such as addition, subtraction, multiplication, division and modular exponentiation. We delve into the detailed quantum implementations of these prominent designs and evaluate their efficiency considering various objectives. We also discuss potential applications of presented arithmetic circuits and suggest future research directions.

quant-ph↗