arXiv subjects

Ziyuan Zhang

Publications and source records attributed to Ziyuan Zhang.

16 recordsLinked to original sources

arXiv2026

Neural Networks Learning the Radon--Nikodym Derivative: Empirical Option Pricing in Incomplete Markets

In incomplete markets, no-arbitrage (NFLVR) guarantees the existence, not the uniqueness, of an equivalent local martingale measure (ELMM): unhedgeable risks (jumps, stochastic volatility) admit a whole family of equivalent measures, and asset dynamics alone cannot pin down the one the market selects. We characterize the identifiability of $Q$ from option data and propose a measure that is identifiable from data yet prices any claim consistently. The key boundary is an ``identification wall'': European options identify only the terminal marginal, while out-of-sample tails and path/joint structure require instruments matched to the priced risk (variance or higher-moment swaps, path-dependent claims). Within this view, minimum-relative-entropy weighted Monte Carlo (WMC) is the optimal baseline for the marginal; we generalize it to a full path-space measure change parameterized by a neural network on physical scenarios---XiNet---which learns $ξ=\mathrm{d}\mathbb{Q}/\mathrm{d}\mathbb{P}$ directly from 8 model-free path features, with option prices as a soft constraint. On European (marginal) pricing XiNet matches but does not surpass WMC, and both are bound by the identification wall out-of-sample. On path-dependent claims the picture reverses: calibrated on the same European surface, per-marginal methods fail structurally (an ATM forward-start is mispriced by $\sim+100\%$), whereas XiNet's single self-consistent measure keeps the bias to $+0.3\%$, beating maximum-entropy WMC ($-24\%$), because $ξ=f_θ(\text{path features})$ captures joint structure Europeans cannot constrain. Identification is thus risk-specific, and XiNet is a single measure that absorbs available instruments and prices all claims consistently.

stat.AP↗

arXiv2026

IntentTester: Intent-Driven Multi-agent Framework for Cross-Library Test Migration

Unit tests capture both functional checks and domain-specific knowledge, but this knowledge remains locked within individual projects and is rarely reused across libraries with overlapping functionality. Existing migration techniques based on structural code mappings (e.g., API signatures) often break down under divergent designs or cross-language settings, resulting in non-executable migrated tests. In this paper, we present IntentTester, a multi-agent framework for intent-driven test reuse. Instead of translating raw code, IntentTester abstracts tests into a language-agnostic Test Description Language (TDL), aligns them with semantically related entities and dependencies in a repository graph, and synthesizes executable tests through LLM-guided reasoning and iterative validation. This design enables cross-library and cross-language migration without manual intervention, producing migrated tests that existing structure-mapping approaches cannot achieve. We evaluate IntentTester on nine open-source projects across three domains (JSON, HTML, and Time) and two languages (Java and Python). IntentTester generates 2,776 syntactically correct tests with 85\% correctness; in comparison, the two baselines achieve 51\% and 43\%. Among them, 2,410 tests executed successfully, yielding a 74\% effectiveness rate. Beyond higher success rates, IntentTester also surfaced previously unknown defects including stack overflows, null dereferences, and parsing inconsistencies, several of which have been acknowledged or patched by maintainers. Our results show that intent-driven migration shifts the focus from code mappings to semantic alignment, allowing practical cross-library and cross-language test reuse while improving test quality and exposing implementation flaws.

cs.SE↗

arXiv2026

Comparing Exploration-Exploitation Strategies of LLMs and Humans: Insights from Standard Multi-armed Bandit Experiments

Large language models (LLMs) are increasingly used to simulate or automate human behavior in complex sequential decision-making settings. A natural question is then whether LLMs exhibit similar decision-making behavior to humans, and can achieve comparable (or superior) performance. In this work, we focus on the exploration-exploitation (E&E) tradeoff, a fundamental aspect of dynamic decision-making under uncertainty. We employ canonical multi-armed bandit (MAB) experiments introduced in the cognitive science and psychiatry literature to conduct a comparative study of the E&E strategies of LLMs, humans, and MAB algorithms. We use interpretable choice models to capture the E&E strategies of the agents and investigate how enabling thinking traces, through both prompting strategies and thinking models, shapes LLM decision-making. We find that enabling thinking in LLMs shifts their behavior toward more human-like behavior, characterized by a mix of random and directed exploration. In a simple stationary setting, thinking-enabled LLMs exhibit similar levels of random and directed exploration compared to humans. However, in more complex, non-stationary environments, LLMs struggle to match human adaptability, particularly in effective directed exploration, despite achieving similar regret in certain scenarios. Our findings highlight both the promise and limits of LLMs as simulators of human behavior and tools for automated decision-making and point to potential areas for improvement.

cs.LG↗

arXiv2026

A Benchmark and Multi-Agent System for Instruction-driven Cinematic Video Compilation

The surging demand for adapting long-form cinematic content into short videos has motivated the need for versatile automatic video compilation systems. However, existing compilation methods are limited to predefined tasks, and the community lacks a comprehensive benchmark to evaluate the cinematic compilation. To address this, we introduce CineBench, the first benchmark for instruction-driven cinematic video compilation, featuring diverse user instructions and high-quality ground-truth compilations annotated by professional editors. To overcome contextual collapse and temporal fragmentation, we present CineAgents, a multi-agent system that reformulates cinematic video compilation into ``design-and-compose'' paradigm. CineAgents performs script reverse-engineering to construct a hierarchical narrative memory to provide multi-level context and employs an iterative narrative planning process that refines a creative blueprint into a final compiled script. Extensive experiments demonstrate that CineAgents significantly outperforms existing methods, generating compilations with superior narrative coherence and logical coherence.

cs.CV↗

arXiv2025

An Engorgio Prompt Makes Large Language Model Babble on

Auto-regressive large language models (LLMs) have yielded impressive performance in many real-world tasks. However, the new paradigm of these LLMs also exposes novel threats. In this paper, we explore their vulnerability to inference cost attacks, where a malicious user crafts Engorgio prompts to intentionally increase the computation cost and latency of the inference process. We design Engorgio, a novel methodology, to efficiently generate adversarial Engorgio prompts to affect the target LLM's service availability. Engorgio has the following two technical contributions. (1) We employ a parameterized distribution to track LLMs' prediction trajectory. (2) Targeting the auto-regressive nature of LLMs' inference process, we propose novel loss functions to stably suppress the appearance of the token, whose occurrence will interrupt the LLM's generation process. We conduct extensive experiments on 13 open-sourced LLMs with parameters ranging from 125M to 30B. The results show that Engorgio prompts can successfully induce LLMs to generate abnormally long outputs (i.e., roughly 2-13$\times$ longer to reach 90%+ of the output length limit) in a white-box scenario and our real-world experiment demonstrates Engergio's threat to LLM service with limited computing resources. The code is released at: https://github.com/jianshuod/Engorgio-prompt.

cs.CR↗

arXiv2025

Building Neutron Stars with the MUSES Calculation Engine

Exploring the equation of state of dense matter is an essential part of interpreting the observable properties of neutron stars. We present here the first results for dense matter in the zero-temperature limit generated by the MUSES Calculation Engine, a composable workflow management system that orchestrates calculation and data processing stages comprising a collection of software modules designed within the MUSES framework. The modules presented in this work calculate equations of state using algorithms spanning three different theories/models: (1) Crust Density Functional Theory, valid starting at low densities, (2) Chiral Effective Field Theory, valid around saturation density, and (3) the Chiral Mean Field model, valid beyond saturation density. Lepton contributions are added through the Lepton module to each equation of state, ensuring charge neutrality and the possibility of $β$-equilibrium. Using the Synthesis module, we match the three equations of state using different thermodynamic variables and different methods. We then couple the complete equation of state to a novel full-general-relativity solver (QLIMR) module that calculates neutron star properties. We find that the matching performed using different thermodynamic variables affects differently the range obtained for neutron star masses and radii (although never beyond a few percent difference). We also investigate the universality of equation of state-independent relations for our matched stars. Finally, for the first time, we use the Flavor Equilibration module to estimate bulk viscosity and flavor relaxation charge fraction and rates (at low temperature) for Chiral Effective Field Theory and the Chiral Mean Field model.

nucl-th↗

arXiv2024

COSMIC: Compress Satellite Images Efficiently via Diffusion Compensation

With the rapidly increasing number of satellites in space and their enhanced capabilities, the amount of earth observation images collected by satellites is exceeding the transmission limits of satellite-to-ground links. Although existing learned image compression solutions achieve remarkable performance by using a sophisticated encoder to extract fruitful features as compression and using a decoder to reconstruct, it is still hard to directly deploy those complex encoders on current satellites' embedded GPUs with limited computing capability and power supply to compress images in orbit. In this paper, we propose COSMIC, a simple yet effective learned compression solution to transmit satellite images. We first design a lightweight encoder (i.e. reducing FLOPs by 2.6~5x) on satellite to achieve a high image compression ratio to save satellite-to-ground links. Then, for reconstructions on the ground, to deal with the feature extraction ability degradation due to simplifying encoders, we propose a diffusion-based model to compensate image details when decoding. Our insight is that satellite's earth observation photos are not just images but indeed multi-modal data with a nature of Text-to-Image pairing since they are collected with rich sensor data (e.g. coordinates, timestamp, etc.) that can be used as the condition for diffusion generation. Extensive experiments show that COSMIC outperforms state-of-the-art baselines on both perceptual and distortion metrics.

eess.IV↗

arXiv2024

Emergence of microphysical bulk viscosity in binary neutron star post-merger dynamics

In nuclear matter in isolated neutron stars, the flavor content (e.g., proton fraction) is subject to weak interactions, establishing flavor ($β$-)equilibrium. However, there can be deviations from this equilibrium during the merger of two neutron stars. We study the resulting out-of-equilibrium dynamics during the collision by incorporating direct and modified Urca processes (in the neutrino-transparent regime) into general-relativistic hydrodynamics simulations with a simplified neutrino transport scheme. We demonstrate how weak-interaction-driven bulk viscosity in post-merger simulations can emerge and assess the bulk viscous dynamics of the resulting flow. We further place limits on the impact on the post-merger gravitational wave strain. Our results show that weak-interaction-driven bulk viscosity can potentially lead to a phase shift of the post-merger gravitational wave spectrum, although the effect is currently on the same level as the numerical errors of our simulation.

astro-ph.HE↗

arXiv2024

Beyond modified Urca: the nucleon width approximation for flavor-changing processes in dense matter

Flavor-changing charged current ("Urca") processes are of central importance in the astrophysics of neutron stars. Standard calculations approximate the Urca rate as the sum of two contributions, direct Urca and modified Urca. Attempts to make modified Urca calculations more accurate have been impeded by an unphysical divergence at the direct Urca threshold density. In this paper we describe a systematically improvable approach where, in the simplest approximation, instead of modified Urca we include an imaginary part of the nucleon mass (nucleon width). The total Urca rate is then obtained via a straightforward generalization of the direct Urca calculation, yielding results that agree with both direct and modified Urca at the densities where those approximations are valid. At low densities, we observe an enhancement of the rate by more than an order of magnitude, with important ramifications for neutron star cooling and other transport properties.

nucl-th↗

arXiv2024

Theoretical and Experimental Constraints for the Equation of State of Dense and Hot Matter

This review aims at providing an extensive discussion of modern constraints relevant for dense and hot strongly interacting matter. It includes theoretical first-principle results from lattice and perturbative QCD, as well as chiral effective field theory results. From the experimental side, it includes heavy-ion collision and low-energy nuclear physics results, as well as observations from neutron stars and their mergers. The validity of different constraints, concerning specific conditions and ranges of applicability, is also provided.

nucl-th↗

arXiv2024

Isospin Equilibration in Neutron Star Mergers

We analyze the isospin equilibration properties of neutrinoless nuclear ($npe$) matter in the temperature and density range that is relevant to neutron star mergers. Our analysis incorporates neutrino-transparency corrections to the isospin (``beta'') equilibrium condition which become noticeable at $T\gtrsim 1\,$MeV. We find that the isospin relaxation rate rises rapidly as temperature rises, and at $T\approx 5\,$MeV it is comparable to the timescale of the density oscillations that occur immediately after the merger. This produces a resonant peak in the bulk viscosity at $T\approx 5\,$MeV, which causes density oscillations to be damped on the timescale of the merger. Our calculations suggest that isospin relaxation dynamics may also be relevant when neutrinos are treated more accurately via neutrino transport schemes.

nucl-th↗

arXiv2023

Beta equilibrium under neutron star merger conditions

We calculate the nonzero-temperature correction to the beta equilibrium condition in nuclear matter under neutron star merger conditions, in the temperature range $1\,$MeV$ < T \lesssim 5\,$MeV. We improve on previous work by using a consistent description of nuclear matter based on the IUF and SFHo relativistic mean field models. This includes using relativistic dispersion relations for the nucleons, which we show is essential in these models. We find that the nonzero-temperature correction can be of order $10$ to $20\,$MeV, and plays an important role in the correct calculation of Urca rates, which can be wrong by factors of $10$ or more if it is neglected.

nucl-th↗

arXiv2023

Aegis: Mitigating Targeted Bit-flip Attacks against Deep Neural Networks

Bit-flip attacks (BFAs) have attracted substantial attention recently, in which an adversary could tamper with a small number of model parameter bits to break the integrity of DNNs. To mitigate such threats, a batch of defense methods are proposed, focusing on the untargeted scenarios. Unfortunately, they either require extra trustworthy applications or make models more vulnerable to targeted BFAs. Countermeasures against targeted BFAs, stealthier and more purposeful by nature, are far from well established. In this work, we propose Aegis, a novel defense method to mitigate targeted BFAs. The core observation is that existing targeted attacks focus on flipping critical bits in certain important layers. Thus, we design a dynamic-exit mechanism to attach extra internal classifiers (ICs) to hidden layers. This mechanism enables input samples to early-exit from different layers, which effectively upsets the adversary's attack plans. Moreover, the dynamic-exit mechanism randomly selects ICs for predictions during each inference to significantly increase the attack cost for the adaptive attacks where all defense mechanisms are transparent to the adversary. We further propose a robustness training strategy to adapt ICs to the attack scenarios by simulating BFAs during the IC training phase, to increase model robustness. Extensive evaluations over four well-known datasets and two popular DNN structures reveal that Aegis could effectively mitigate different state-of-the-art targeted attacks, reducing attack success rate by 5-10$\times$, significantly outperforming existing defense methods.

cs.CR↗

arXiv2022

Mind Your Heart: Stealthy Backdoor Attack on Dynamic Deep Neural Network in Edge Computing

Transforming off-the-shelf deep neural network (DNN) models into dynamic multi-exit architectures can achieve inference and transmission efficiency by fragmenting and distributing a large DNN model in edge computing scenarios (e.g., edge devices and cloud servers). In this paper, we propose a novel backdoor attack specifically on the dynamic multi-exit DNN models. Particularly, we inject a backdoor by poisoning one DNN model's shallow hidden layers targeting not this vanilla DNN model but only its dynamically deployed multi-exit architectures. Our backdoored vanilla model behaves normally on performance and cannot be activated even with the correct trigger. However, the backdoor will be activated when the victims acquire this model and transform it into a dynamic multi-exit architecture at their deployment. We conduct extensive experiments to prove the effectiveness of our attack on three structures (ResNet-56, VGG-16, and MobileNet) with four datasets (CIFAR-10, SVHN, GTSRB, and Tiny-ImageNet) and our backdoor is stealthy to evade multiple state-of-the-art backdoor detection or removal methods.

cs.CR↗

arXiv2019

On Learning Disentangled Representations for Gait Recognition

Gait, the walking pattern of individuals, is one of the important biometrics modalities. Most of the existing gait recognition methods take silhouettes or articulated body models as gait features. These methods suffer from degraded recognition performance when handling confounding variables, such as clothing, carrying and viewing angle. To remedy this issue, we propose a novel AutoEncoder framework, GaitNet, to explicitly disentangle appearance, canonical and pose features from RGB imagery. The LSTM integrates pose features over time as a dynamic gait feature while canonical features are averaged as a static gait feature. Both of them are utilized as classification features. In addition, we collect a Frontal-View Gait (FVG) dataset to focus on gait recognition from frontal-view walking, which is a challenging problem since it contains minimal gait cues compared to other views. FVG also includes other important variations, e.g., walking speed, carrying, and clothing. With extensive experiments on CASIA-B, USF, and FVG datasets, our method demonstrates superior performance to the SOTA quantitatively, the ability of feature disentanglement qualitatively, and promising computational efficiency. We further compare our GaitNet with state-of-the-art face recognition to demonstrate the advantages of gait biometrics identification under certain scenarios, e.g., long distance/lower resolutions, cross viewing angles.

cs.CV↗

arXiv2019

Gait Recognition via Disentangled Representation Learning

Gait, the walking pattern of individuals, is one of the most important biometrics modalities. Most of the existing gait recognition methods take silhouettes or articulated body models as the gait features. These methods suffer from degraded recognition performance when handling confounding variables, such as clothing, carrying and view angle. To remedy this issue, we propose a novel AutoEncoder framework to explicitly disentangle pose and appearance features from RGB imagery and the LSTM-based integration of pose features over time produces the gait feature. In addition, we collect a Frontal-View Gait (FVG) dataset to focus on gait recognition from frontal-view walking, which is a challenging problem since it contains minimal gait cues compared to other views. FVG also includes other important variations, e.g., walking speed, carrying, and clothing. With extensive experiments on CASIA-B, USF and FVG datasets, our method demonstrates superior performance to the state of the arts quantitatively, the ability of feature disentanglement qualitatively, and promising computational efficiency.

cs.CV↗