arXiv subjects

Haoxin Wang

Publications and source records attributed to Haoxin Wang.

At least 19 recordsLinked to original sources

arXiv2026

How Do Prompt Variations Affect Energy Consumption in On-Device LLMs?

Large language models (LLMs) are increasingly deployed on mobile devices, making energy efficiency a key deployment constraint, yet the energy impact of prompt design remains underexplored. This paper aims to understand how two prompt properties, cognitive load and phrasing pattern, shape the energy behavior of on-device LLM inference. We conduct a broad empirical study covering prompt properties, datasets, models, and devices, with phase-level profiling that separates prefill and decode energy. We find that cognitive load primarily affects the energy cost per token, while phrasing pattern affects energy largely through token usage. Our energy-quality analysis further shows that prompt design reshapes the attainable frontier differently across models, highlighting the need for model-aware prompt design in energy-efficient on-device LLM inference. Code, datasets, and scripts are available at https://amai-gsu.github.io/PromptProperty/.

cs.CL↗

arXiv2026

Hierarchical Federated Transfer Learning in Digital Twin-Based Vehicular Networks

In recent research on the Digital Twin-based Vehicular Ad hoc Network(DT-VANET), Federated Learning (FL) has shown its ability to provide data privacy. However, Federated learning struggles to adequately train a global model when confronted with data heterogeneity and data sparsity among vehicles, which ensure suboptimal accuracy in making precise predictions for different vehicle types. To address these challenges, this paper combines Federated Transfer Learning (FTL) to conduct vehicle clustering related to types of vehicles and proposes a novel Hierarchical Federated Transfer Learning (HFTL). We construct a framework for DT-VANET, along with two algorithms designed for cloud server model updates and intra-cluster federated transfer learning, to improve the accuracy of the global model. In addition, we developed a data quality score-based mechanism to prevent the global model from being affected by malicious vehicles. Lastly, detailed experiments on real-world datasets are conducted, considering different performance metrics that verify the effectiveness and efficiency of our algorithm.

cs.LG↗

arXiv2026

EpiFormer: Learning Antigen-Antibody Interactions for Epitope Prediction via Geometric Deep Learning

Antibodies neutralize foreign antigens by binding to specific surface regions called epitopes. Computational epitope prediction is critical for understanding immune recognition and guiding antibody engineering. However, existing methods face three fundamental challenges: antibody-aware models encode each chain independently and combine them only at a late stage, failing to capture co-dependent structural features that define binding interfaces, whereas severe class imbalance and scarcity of known antibody-antigen complexes render standard training objectives ineffective. We propose EpiFormer, a general encoder-decoder framework that addresses these challenges jointly. Our key design principle is interleaved cross-attention within GNN encoding layers, enabling bidirectional antigen-antibody information flow throughout representation learning rather than only at the output. This early-fusion principle is backbone-agnostic, providing consistent gains across GNN architectures from simple GCNs to equivariant models. We further show that sparsity-aware objectives are effective when paired with early-fusion architectures for the epitope prediction task. EpiFormer improves over the previous best method by over 40% in F1 score on standard benchmarks, demonstrating generalizability and cross-dataset transferability. Notably, EpiFormer discovers known biological principles as emergent behaviors of end-to-end training, where the learned cross-attention gates favor antigen-to-antibody information flow, consistent with the asymmetric roles of the two chains at the binding interface, and the model's preference for geometric over evolutionary features aligns with the established finding that epitope residues are not evolutionarily conserved. The source code is available at: https://github.com/mansoor181/epiformer.git

q-bio.QM↗

arXiv2026

Keyword Matters: Unveiling the Energy Sensitivity of On-Device LLM Prompting

Large Language Models (LLMs) are increasingly deployed on mobile and embedded devices to improve privacy and reduce network latency. Yet on-device inference faces a fundamental constraint: high energy consumption on battery-powered, resource-limited hardware. While model compression and runtime acceleration have been widely studied, the effect of \emph{prompt design} on energy efficiency remains underexplored. This paper presents an empirical study of the relationship between prompt wording and energy consumption for on-device LLMs. Using real power measurements collected on a smartphone, we quantify how linguistic features, particularly imperative keywords and instruction structure, affect decoding length and total energy. Our results show consistent energy differences across verbs and tasks, indicating that prompt engineering is a lightweight lever for improving energy efficiency.

cs.AI↗

arXiv2026

Deformba: Vision State Space Model with Adaptive State Fusion

State Space Models (SSMs) have emerged as a powerful and efficient alternative to Transformers, demonstrating linear-time complexity and exceptional sequence modeling capabilities. However, their application to vision tasks remains challenging. First, existing vision SSMs largely depend on manually designed fixed scanning methods to flatten image patches into sequences, which imposes predefined geometric structures and increases the complexity. Second, the broader adoption of vision SSMs is hindered in domains that require query-based interactions between distinct information streams. This is a result of the inherently causal and self-referential nature of SSMs designed for 1D sequence modeling tasks. This fusion mechanism is indispensable for critical perception tasks such as multi-view 3D fusion. To address these limitations, we propose Deformba, a context adaptive method that dynamically augments the spatial structural information while maintaining the linear complexity of SSMs. Deformba also allows multi-modal fusion like cross attention. To demonstrate the effectiveness and general applicability of Deformba, we test its performance on general 2D vision tasks such as image classification, object detection, and segmentation, as well as 3D vision tasks like BEV perception. Extensive experiments show that Deformba achieves strong performance across various visual perception benchmarks.

cs.CV↗

arXiv2026

Transferable Latency Prediction for Fast LLM Screening on Heterogeneous Edge Devices

Accurate latency prediction is critical for deploying large language models (LLMs) on heterogeneous edge devices, where inference latency is affected by model architecture, prompt behavior, runtime backend, hardware utilization, dynamic voltage and frequency scaling (DVFS), and thermal variation. This paper presents a runtime-aware latency prediction framework for deployment-oriented LLM selection. The framework represents each inference request as a hardware-runtime-model-prompt configuration, separates inference into prefill and decode phases, and adaptively fuses static descriptors with dynamic hardware telemetry through a gated prediction model. We evaluate the framework using Pixel mobile devices and validate the profiling pipeline on Jetson Nano, Orange Pi 5 Pro, and an RTX 3090-class GPU platform. On Pixel 8, the full predictor improves total-latency R-squared from 0.953 to 0.960 and decode-latency R-squared from 0.957 to 0.973 over a static-only baseline. On Pixel 8 Pro, it improves prefill-latency R-squared from -1.383 to 0.966. For cross-device transfer, calibration improves Pixel 8 Pro to Pixel 8 total-latency R-squared from -0.974 to 0.940 and decode-latency R-squared from -1.085 to 0.927. Heterogeneous profiling further shows that latency is highly device- and runtime-dependent: the same SmolLM2 model family reaches 8.42 tokens/s on Orange Pi 5 Pro but 64.38 tokens/s on an RTX 3090-class GPU. These results demonstrate that runtime-aware prediction with lightweight calibration can reduce profiling cost and support latency-aware LLM deployment across heterogeneous edge platforms.

cs.AI↗

arXiv2025

Topological Magnetic Phases and Magnon-Phonon Hybridization in the Presence of Strong Dzyaloshinskii-Moriya Interaction

In recent years, the interplay between quantum magnetism and topology has attracted growing interest, both for its fundamental importance and its technological potential. Topological magnons, quantized spin excitations with nontrivial band topology, hold particular promise for spintronics, offering routes to robust, low-dissipation devices for next-generation information processing and storage. While topological magnons in honeycomb ferromagnets with weak next-nearest-neighbor Dzyaloshinskii-Moriya interactions (DMI) have been extensively investigated, the strong-DMI regime remains largely unexplored. In this work, we examine topological magnetic phases and magnon-phonon hybridization in a two-dimensional magnetic system with strong DMI. We show that strong DMI drives a transition from a ferromagnetic ground state to a 120$^\circ$ noncollinear order. An additional Zeeman field further induces noncoplanar spin textures, giving rise to a diverse set of topological phases. We demonstrate that these topological phases can be directly probed through the anomalous thermal Hall effect. Finally, we find that the spin-spin interactions in the strong-$D$ phase enable magnon-phonon coupling that yields hybridized topological bands, whereas such coupling vanishes in the weak-$D$ phase.

cond-mat.mes-hall↗

arXiv2025

PlatformX: An End-to-End Transferable Platform for Energy-Efficient Neural Architecture Search

Hardware-Aware Neural Architecture Search (HW-NAS) has emerged as a powerful tool for designing efficient deep neural networks (DNNs) tailored to edge devices. However, existing methods remain largely impractical for real-world deployment due to their high time cost, extensive manual profiling, and poor scalability across diverse hardware platforms with complex, device-specific energy behavior. In this paper, we present PlatformX, a fully automated and transferable HW-NAS framework designed to overcome these limitations. PlatformX integrates four key components: (i) an energy-driven search space that expands conventional NAS design by incorporating energy-critical configurations, enabling exploration of high-efficiency architectures; (ii) a transferable kernel-level energy predictor across devices and incrementally refined with minimal on-device samples; (iii) a Pareto-based multi-objective search algorithm that balances energy and accuracy to identify optimal trade-offs; and (iv) a high-resolution runtime energy profiling system that automates on-device power measurement using external monitors without human intervention. We evaluate PlatformX across multiple mobile platforms, showing that it significantly reduces search overhead while preserving accuracy and energy fidelity. It identifies models with up to 0.94 accuracy or as little as 0.16 mJ per inference, both outperforming MobileNet-V2 in accuracy and efficiency. Code and tutorials are available at github.com/amai-gsu/PlatformX.

cs.LG↗

arXiv2025

lm-Meter: Unveiling Runtime Inference Latency for On-Device Language Models

Large Language Models (LLMs) are increasingly integrated into everyday applications, but their prevalent cloud-based deployment raises growing concerns around data privacy and long-term sustainability. Running LLMs locally on mobile and edge devices (on-device LLMs) offers the promise of enhanced privacy, reliability, and reduced communication costs. However, realizing this vision remains challenging due to substantial memory and compute demands, as well as limited visibility into performance-efficiency trade-offs on resource-constrained hardware. We propose lm-Meter, the first lightweight, online latency profiler tailored for on-device LLM inference. lm-Meter captures fine-grained, real-time latency at both phase (e.g., embedding, prefill, decode, softmax, sampling) and kernel levels without auxiliary devices. We implement lm-Meter on commercial mobile platforms and demonstrate its high profiling accuracy with minimal system overhead, e.g., only 2.58% throughput reduction in prefill and 0.99% in decode under the most constrained Powersave governor. Leveraging lm-Meter, we conduct comprehensive empirical studies revealing phase- and kernel-level bottlenecks in on-device LLM inference, quantifying accuracy-efficiency trade-offs, and identifying systematic optimization opportunities. lm-Meter provides unprecedented visibility into the runtime behavior of LLMs on constrained platforms, laying the foundation for informed optimization and accelerating the democratization of on-device LLM systems. Code and tutorials are available at https://github.com/amai-gsu/LM-Meter.

cs.LG↗

arXiv2025

Theoretical Analysis on the Ability of Particle Swarm Optimization to Escape from Local Optimum

As one of the most prominent swarm intelligence algorithms, particle swarm optimization (PSO) has been extensively applied to solve global optimization problems. The theoretical analysis on the ability of PSO to escape from local optimum (LO) is deeply intertwined with its global optimization performance, making it highly significant to investigate the underlying mechanisms that enable PSO to escape from LO and achieve global optimization. In this study, a rigorous mathematical analysis is presented to explore the ability and mechanisms of PSO to escape from LO. By modelling each search agent as a Markov chain, the probability of PSO escaping from LO within a finite number of iterations is calculated, and the behavior of agents prior to escaping from LO is analyzed. Through this theoretical analysis, two necessary and sufficient conditions are proposed to ensure that PSO escapes from LO in finite iterations with a probability 1: (a) the personal best (pbest) and global best (gbest) positions stagnate around different local optima, and (b) the inertial weight in PSO is equal to one. Upon these conditions, two potential behaviors of agents, namely inertial motion and oscillation, are proposed, and a state transition chain of the agent toward the global optimum is constructed. Additionally, the theoretical analysis reveals that increasing the distance between pbest and gbest, as well as the values of the learning factors in PSO, enhances the speed at which PSO escapes from LO. Finally, the conclusions drawn from the theoretical analysis are corroborated through numerical experiments.

math.OC↗

arXiv2025

BadTime: An Effective Backdoor Attack on Multivariate Long-Term Time Series Forecasting

Multivariate long-term time series forecasting (MLTSF) models are increasingly deployed in critical domains such as climate, finance, and transportation. Despite their growing importance, the security of MLTSF models against backdoor attacks remains entirely unexplored. To bridge this gap, we propose BadTime, the first effective backdoor attack tailored for MLTSF. BadTime can manipulate hundreds of future predictions toward a target pattern by injecting a subtle trigger. BadTime addresses two key challenges that arise uniquely in MLTSF: (i) the rapid dilution of local triggers over long horizons, and (ii) the extreme sparsity of backdoor signals under stealth constraints. To counter dilution, BadTime leverages inter-variable correlations, temporal lags, and data-driven initialization to design a distributed, lag-aware trigger that ensures effective influence over long-range forecasts. To overcome sparsity, it introduces a hybrid strategy to select valuable poisoned samples and a decoupled backdoor training objective that adaptively adjusts the model's focus on the sparse backdoor signal, ensuring reliable learning at a poisoning rate as low as 1%. Extensive experiments show that BadTime significantly outperforms state-of-the-art (SOTA) backdoor attacks on time series forecasting by extending the attackable horizon from at most 12 timesteps to 720 timesteps (a 60-fold improvement), reducing MAE by over 50% on target variables, and boosting stealthiness by more than 3-fold under anomaly detection.

cs.CR↗

arXiv2025

ECom-Bench: Can LLM Agent Resolve Real-World E-commerce Customer Support Issues?

In this paper, we introduce ECom-Bench, the first benchmark framework for evaluating LLM agent with multimodal capabilities in the e-commerce customer support domain. ECom-Bench features dynamic user simulation based on persona information collected from real e-commerce customer interactions and a realistic task dataset derived from authentic e-commerce dialogues. These tasks, covering a wide range of business scenarios, are designed to reflect real-world complexities, making ECom-Bench highly challenging. For instance, even advanced models like GPT-4o achieve only a 10-20% pass^3 metric in our benchmark, highlighting the substantial difficulties posed by complex e-commerce scenarios. The code and data have been made publicly available at https://github.com/XiaoduoAILab/ECom-Bench to facilitate further research and development in this domain.

cs.CL↗

arXiv2025

MindFlow: Revolutionizing E-commerce Customer Support with Multimodal LLM Agents

Recent advances in large language models (LLMs) have enabled new applications in e-commerce customer service. However, their capabilities remain constrained in complex, multimodal scenarios. We present MindFlow, the first open-source multimodal LLM agent tailored for e-commerce. Built on the CoALA framework, it integrates memory, decision-making, and action modules, and adopts a modular "MLLM-as-Tool" strategy for effect visual-textual reasoning. Evaluated via online A/B testing and simulation-based ablation, MindFlow demonstrates substantial gains in handling complex queries, improving user satisfaction, and reducing operational costs, with a 93.53% relative improvement observed in real-world deployments.

cs.CL↗

arXiv2025

Symmetry-protected electronic metastability in an optically driven cuprate ladder

Optically excited quantum materials exhibit nonequilibrium states with remarkable emergent properties, but these phenomena are usually transient, decaying on picosecond timescales and limiting practical applications. Advancing the design and control of nonequilibrium phases requires the development of targeted strategies to achieve long-lived, metastable phases. Here, we report the discovery of symmetry-protected electronic metastability in the model cuprate ladder Sr$_{14}$Cu$_{24}$O$_{41}$. Using femtosecond resonant x-ray scattering and spectroscopy, we show that this metastability is driven by a transfer of holes from chain-like charge reservoirs into the ladders. This ultrafast charge redistribution arises from the optical dressing and activation of a hopping pathway that is forbidden by symmetry at equilibrium. Relaxation back to the ground state is hence suppressed after the pump coherence dissipates. Our findings highlight how dressing materials with electromagnetic fields can dynamically activate terms in the electronic Hamiltonian, and provide a rational design strategy for nonequilibrium phases of matter.

cond-mat.str-el↗

arXiv2025

MamBEV: Enabling State Space Models to Learn Birds-Eye-View Representations

3D visual perception tasks, such as 3D detection from multi-camera images, are essential components of autonomous driving and assistance systems. However, designing computationally efficient methods remains a significant challenge. In this paper, we propose a Mamba-based framework called MamBEV, which learns unified Bird's Eye View (BEV) representations using linear spatio-temporal SSM-based attention. This approach supports multiple 3D perception tasks with significantly improved computational and memory efficiency. Furthermore, we introduce SSM based cross-attention, analogous to standard cross attention, where BEV query representations can interact with relevant image features. Extensive experiments demonstrate MamBEV's promising performance across diverse visual perception metrics, highlighting its advantages in input scaling efficiency compared to existing benchmark models.

cs.CV↗

arXiv2025

The Gradient Puppeteer: Adversarial Domination in Gradient Leakage Attacks through Model Poisoning

In Federated Learning (FL), clients share gradients with a central server while keeping their data local. However, malicious servers could deliberately manipulate the models to reconstruct clients' data from shared gradients, posing significant privacy risks. Although such active gradient leakage attacks (AGLAs) have been widely studied, they suffer from two severe limitations: (i) coverage: no existing AGLAs can reconstruct all samples in a batch from the shared gradients; (ii) stealthiness: no existing AGLAs can evade principled checks of clients. In this paper, we address these limitations with two core contributions. First, we introduce a new theoretical analysis approach, which uniformly models AGLAs as backdoor poisoning. This analysis approach reveals that the core principle of AGLAs is to bias the gradient space to prioritize the reconstruction of a small subset of samples while sacrificing the majority, which theoretically explains the above limitations of existing AGLAs. Second, we propose Enhanced Gradient Global Vulnerability (EGGV), the first AGLA that achieves complete attack coverage while evading client-side detection. In particular, EGGV employs a gradient projector and a jointly optimized discriminator to assess gradient vulnerability, steering the gradient space toward the point most prone to data leakage. Extensive experiments show that EGGV achieves complete attack coverage and surpasses state-of-the-art (SOTA) with at least a 43% increase in reconstruction quality (PSNR) and a 45% improvement in stealthiness (D-SNR).

cs.CR↗

arXiv2025

GreenAuto: An Automated Platform for Sustainable AI Model Design on Edge Devices

We present GreenAuto, an end-to-end automated platform designed for sustainable AI model exploration, generation, deployment, and evaluation. GreenAuto employs a Pareto front-based search method within an expanded neural architecture search (NAS) space, guided by gradient descent to optimize model exploration. Pre-trained kernel-level energy predictors estimate energy consumption across all models, providing a global view that directs the search toward more sustainable solutions. By automating performance measurements and iteratively refining the search process, GreenAuto demonstrates the efficient identification of sustainable AI models without the need for human intervention.

cs.LG↗

arXiv2024

CarbonCP: Carbon-Aware DNN Partitioning with Conformal Prediction for Sustainable Edge Intelligence

This paper presents a solution to address carbon emission mitigation for end-to-end edge computing systems, including the computing at battery-powered edge devices and servers, as well as the communications between them. We design and implement, CarbonCP, a context-adaptive, carbon-aware, and uncertainty-aware AI inference framework built upon conformal prediction theory, which balances operational carbon emissions, end-to-end latency, and battery consumption of edge devices through DNN partitioning under varying system processing contexts and carbon intensity. Our experimental results demonstrate that CarbonCP is effective in substantially reducing operational carbon emissions, up to 58.8%, while maintaining key user-centric performance metrics with only 9.9% error rate.

cs.NI↗