arXiv subjects

Hai Huang

Publications and source records attributed to Hai Huang.

At least 19 recordsLinked to original sources

arXiv2026

Massless scalar scattering by Kerr-Bertotti-Robinson black holes:transparent-end channels and superradiance

We formulate and numerically solve the real-frequency scattering problem for a neutral, minimally coupled, massless scalar on the rotating Kerr--Bertotti--Robinson (Kerr--BR) black-hole geometry, using a specified transparent boundary condition at the coordinate end. Since the Maxwell stress tensor of the background is traceless, the minimally coupled wave equation reduces to the four-dimensional conformal wave equation, enabling exact Carter-like separation after scaling the scalar field by the conformal factor. Unlike the asymptotically flat case, the coordinate end $r\to\infty$ lies at a finite tortoise distance. We show that the resulting reflection data are conditional on this boundary prescription rather than defining a unique, observer-independent cross section. Within the transparent-end model,open-channel superradiance is governed by a double-gate mechanism requiring both the local horizon condition and the outer propagation condition $q_\infty^2>0$. At the benchmark spin $a/M=0.9$, the co-rotating dipole amplification decreases as the external magnetic field increases. Crucially, the field narrows and closes the open superradiant window at $BM\simeq0.243$. Near the propagation threshold, the amplification coefficient vanishes linearly with the outer wave number.

gr-qc↗

arXiv2026

Scattering and absorption of a charged massive scalar field by a Reissner-Nordstr\"om black hole surrounded by perfect fluid dark matter

We study the scattering of charged massive particles impinging on a Reissner-Nordstr\"{o}m (RN) black hole immersed in perfect fluid dark matter (PFDM). We obtain an approximation of absorption cross section in the low-frequency regime via the matching method. In the high-frequency regime, we derive the weak-field deflection angle up to second order. The numerical results are in excellent agreement with classical approximation and glory scattering. The effects of dark matter, particle charge, and mass upon scattering and absorption are examined in detail. The results show that as the dark matter parameter $\lambda$ increases, the absorption cross section of the black hole is strongly suppressed, and its high-frequency limit depends only on the black hole charge $Q$ and $\lambda$. The scattering cross section also decreases overall. In the superradiant regime, the amplification factor of the PFDM black hole is much larger than that of the RN black hole. Finally, we discuss the behavior of the absorption cross section as $\omega/m\rightarrow1$, as well as the scattering cross section at small scattering angles.

gr-qc↗

arXiv2026

Recovery of tunable bound states in the continuum

Tunable bound states in the continuum (BICs) in photonic crystal slabs are highly sensitive to substrate-induced mirror-symmetry breaking and typically degrade into finite-$Q$ quasi-BICs in realistic integrated platforms. Here we show that such degradation can be deterministically reversed. Using temporal coupled-mode theory and full-wave simulations, we demonstrate that the radiation channel opened by the substrate can be exactly cancelled by introducing a second, independent odd-parity perturbation inside the slab. This dual-asymmetry strategy restores the singularity of the radiation matrix and thereby recovers a tunable BIC in a substrate-supported photonic crystal slab. The recovered state regains both the polarization vortex and the characteristic $Q\propto \Delta k^{-2}$ scaling. The recovery points further follow a linear relation in the two-asymmetry parameter space, revealing a simple mode-dependent compensation law. The same mechanism also restores merging-BIC configurations, showing that it applies not only to isolated tunable BICs but also to higher-order topological resonance states built from them. Our results establish a practical route for preserving tunable topological resonances in substrate-supported nanophotonic systems.

physics.optics↗

arXiv2026

Emerging trends in Cislunar Space for Lunar Science Exploration and Space Robotics aiding Human Spaceflight Safety

In recent years, the Moon has emerged as an unparalleled extraterrestrial testbed for advancing cuttingedge technological and scientific research critical to enabling sustained human presence on its surface and supporting future interplanetary exploration. This study identifies and investigates two pivotal research domains with substantial transformative potential for accelerating humanity interplanetary aspirations. First is Lunar Science Exploration with Artificial Intelligence and Space Robotics which focusses on AI and Space Robotics redefining the frontiers of space exploration. Second being Space Robotics aid in manned spaceflight to the Moon serving as critical assets for pre-deployment infrastructure development, In-Situ Resource Utilization, surface operations support, and astronaut safety assurance. By integrating autonomy, machine learning, and realtime sensor fusion, space robotics not only augment human capabilities but also serve as force multipliers in achieving sustainable lunar exploration, paving the way for future crewed missions to Mars and beyond.

cs.RO↗

arXiv2026

Semantic Tube Prediction: Beating LLM Data Efficiency with JEPA

Large Language Models (LLMs) obey consistent scaling laws -- empirical power-law fits that predict how loss decreases with compute, data, and parameters. While predictive, these laws are descriptive rather than prescriptive: they characterize typical training, not optimal training. Surprisingly few works have successfully challenged the data-efficiency bounds implied by these laws -- which is our primary focus. To that end, we introduce the Geodesic Hypothesis, positing that token sequences trace geodesics on a smooth semantic manifold and are therefore locally linear. Building on this principle, we propose a novel Semantic Tube Prediction (STP) task, a JEPA-style regularizer that confines hidden-state trajectories to a tubular neighborhood of the geodesic. STP generalizes JEPA to language without requiring explicit multi-view augmentations. We show this constraint improves signal-to-noise ratio, and consequently preserves diversity by preventing trajectory collisions during inference. Empirically, STP allows LLMs to match baseline accuracy with 16$\times$ less training data on the NL-RX-SYNTH dataset, directly violating the data term of Chinchilla-style scaling laws and demonstrating that principled geometric priors can surpass brute-force scaling. Code is available at https://github.com/galilai-group/llm-jepa#stp.

cs.LG↗

arXiv2026

Sparse Models, Sparse Safety: Unsafe Routes in Mixture-of-Experts LLMs

By introducing routers to selectively activate experts in Transformer layers, the mixture-of-experts (MoE) architecture significantly reduces computational costs in large language models (LLMs) while maintaining competitive performance, especially for models with massive parameters. However, prior work has largely focused on utility and efficiency, leaving the safety risks associated with this sparse architecture underexplored. In this work, we show that the safety of MoE LLMs is as sparse as their architecture by discovering unsafe routes: routing configurations that, once activated, convert safe outputs into harmful ones. Specifically, we first introduce the Router Safety importance score (RoSais) to quantify the safety criticality of each layer's router. Manipulation of only the high-RoSais router(s) can flip the default route into an unsafe one. For instance, on JailbreakBench, masking 5 routers in DeepSeek-V2-Lite increases attack success rate (ASR) by over 4$\times$ to 0.79, highlighting an inherent risk that router manipulation may naturally occur in MoE LLMs. We further propose a Fine-grained token-layer-wise Stochastic Optimization framework to discover more concrete Unsafe Routes (F-SOUR), which explicitly considers the sequentiality and dynamics of input tokens. Across four representative MoE LLM families, F-SOUR achieves an average ASR of 0.90 and 0.98 on JailbreakBench and AdvBench, respectively. Finally, we outline defensive perspectives, including safety-aware route disabling and router training, as promising directions to safeguard MoE LLMs. We hope our work can inform future red-teaming and safeguarding of MoE LLMs. Our code is provided in https://github.com/TrustAIRLab/UnsafeMoE.

cs.LG↗

arXiv2026

PrAg-PO: Prompt Augmented Policy Optimization for Robust and Diverse Mathematical Reasoning

Reinforcement learning algorithms such as group-relative policy optimization (GRPO) have shown strong potential for improving the mathematical reasoning capabilities of large language models. While a growing body of work seeks to improve training entropy, rollout diversity, and exploration, most existing methods still train models with a single fixed reasoning prompt or template, which can encourage prompt-specific overfitting and unstable training dynamics. In this work, we introduce Prompt Augmented Policy Optimization (PrAg-PO), a simple policy optimization method that mixes prompt templates with template-specific format rewards during training. By encouraging models to generate reasoning traces under diverse instructions and output formats, PrAg-PO increases rollout diversity and improves robustness. Compared with GRPO and DAPO, PrAg-PO achieves significantly higher reasoning accuracy while mitigating premature training collapse. Empirically, experiments on DeepSeek-R1-Distill-Qwen-1.5B, Qwen2.5-Math-1.5B, and Qwen3-1.7B show that PrAg-PO consistently outperforms strong baselines and achieves competitive performance against recent methods on mathematics benchmarks, using only a fixed MATH Level 3-5 training set of 8.5K problems. The code and model checkpoints are available at https://github.com/wenquanlu/PrAg-PO.

cs.LG↗

arXiv2025

EquaCode: A Multi-Strategy Jailbreak Approach for Large Language Models via Equation Solving and Code Completion

Large language models (LLMs), such as ChatGPT, have achieved remarkable success across a wide range of fields. However, their trustworthiness remains a significant concern, as they are still susceptible to jailbreak attacks aimed at eliciting inappropriate or harmful responses. However, existing jailbreak attacks mainly operate at the natural language level and rely on a single attack strategy, limiting their effectiveness in comprehensively assessing LLM robustness. In this paper, we propose Equacode, a novel multi-strategy jailbreak approach for large language models via equation-solving and code completion. This approach transforms malicious intent into a mathematical problem and then requires the LLM to solve it using code, leveraging the complexity of cross-domain tasks to divert the model's focus toward task completion rather than safety constraints. Experimental results show that Equacode achieves an average success rate of 91.19% on the GPT series and 98.65% across 3 state-of-the-art LLMs, all with only a single query. Further, ablation experiments demonstrate that EquaCode outperforms either the mathematical equation module or the code module alone. This suggests a strong synergistic effect, thereby demonstrating that multi-strategy approach yields results greater than the sum of its parts.

cs.CR↗

arXiv2025

Rewritable Complementary Nanoelectronics Enabled by Electron-Beam Programmable Ambipolar Doping

The ability to reversibly and site-selectively tune ambipolar doping in a single semiconductor is crucial for reconfigurable electronics beyond silicon, but remains highly challenging. Here, we present a rewritable architecture based on electron-beam programmable field-effect transistors (FETs). Using WSe$_2$ as a model system, we demonstrate electron-beam-induced doping that enables reversible, precisely controlled carrier modulation exceeding $10^{13}$ cm$^{-2}$. The in-situ writing, erasing, and rewriting of ambipolar doping of nanoscale patterns was directly visualized by scanning microwave impedance microscopy. This mask-free, lithography-compatible approach can achieve precise band engineering within individual channels, yielding near-ideal subthreshold swings (~ 60 mV/dec) and finely tunable threshold voltages for both carrier types without specialized contact engineering. These capabilities allow on-demand realization of high performance logic, including CMOS inverters with high voltage gains and low power consumption, as well as NAND-to-NOR transitions on the same device via direct polarity rewriting. Our platform offers a scalable and versatile route for rapid prototyping of complementary electronics.

cond-mat.mtrl-sci↗

arXiv2025

UMDAM: A Unified Data Layout and DRAM Address Mapping for Heterogenous NPU-PIM

Large Language Models (LLMs) are increasingly deployed on edge devices with Neural Processing Units (NPUs), yet the decode phase remains memory-intensive, limiting performance. Processing-in-Memory (PIM) offers a promising solution, but co-executing NPU-PIM systems face challenges such as data layout mismatches, bandwidth loss, and redundant storage. To address these issues, we propose UMDAM, a unified memory-affinity data layout and DRAM address mapping scheme tailored for NPU-PIM co-execution. UMDAM employs a column-major, tile-based layout and a configurable DRAM mapping strategy to ensure compatibility with NPU computation while maximizing PIM efficiency -- without introducing extra memory overhead or bandwidth loss. Comprehensive evaluations on OPT models demonstrate that UMDAM reduces time-to-first-token (TTFT) by up to 3.0x and time-to-last-token (TTLT) by 2.18x, significantly improving end-to-end LLM inference efficiency on edge devices.

cs.DC↗

arXiv2025

Unifying and Enhancing Graph Transformers via a Hierarchical Mask Framework

Graph Transformers (GTs) have emerged as a powerful paradigm for graph representation learning due to their ability to model diverse node interactions. However, existing GTs often rely on intricate architectural designs tailored to specific interactions, limiting their flexibility. To address this, we propose a unified hierarchical mask framework that reveals an underlying equivalence between model architecture and attention mask construction. This framework enables a consistent modeling paradigm by capturing diverse interactions through carefully designed attention masks. Theoretical analysis under this framework demonstrates that the probability of correct classification positively correlates with the receptive field size and label consistency, leading to a fundamental design principle: an effective attention mask should ensure both a sufficiently large receptive field and a high level of label consistency. While no single existing mask satisfies this principle across all scenarios, our analysis reveals that hierarchical masks offer complementary strengths, motivating their effective integration. Then, we introduce M3Dphormer, a Mixture-of-Experts-based Graph Transformer with Multi-Level Masking and Dual Attention Computation. M3Dphormer incorporates three theoretically grounded hierarchical masks and employs a bi-level expert routing mechanism to adaptively integrate multi-level interaction information. To ensure scalability, we further introduce a dual attention computation scheme that dynamically switches between dense and sparse modes based on local mask sparsity. Extensive experiments across multiple benchmarks demonstrate that M3Dphormer achieves state-of-the-art performance, validating the effectiveness of our unified framework and model design.

cs.CV↗

arXiv2025

Synergistic Modality-and-Slice Memory Framework for Cross-Modal 3D Brain Tumor Segmentation

The 3D multi-modal brain tumor segmentation is critical to multi-modal healthcare, and it requires accurate identification of distinct internal anatomical subregions. While the recent prompt-based segmentation paradigms enable interactive experiences for clinicians, existing methods ignore cross-modal correlations and rely on labor-intensive category-specific prompts, limiting their applicability in real-world scenarios. To address these issues, we propose the MSM-Seg, a synergistic framework for multi-modal brain tumor segmentation. The MSM-Seg introduces a dual-memory segmentation paradigm that synergistically integrates multi-modal and inter-slice information with an efficient category-agnostic prompt for brain tumor understanding. To this end, we first devise a modality-and-slice memory attention (MSMA) to exploit the complex cross-modal correlations and spatial-slice dependencies among the input scans. \cz{Then, we propose a multi-scale category-agnostic prompt encoder (MCP-Encoder) to provide whole tumor region guidance for decoding.} Moreover, we devise a modality-adaptive fusion decoder (MF-Decoder) that leverages the complementary decoding information across different modalities to improve segmentation accuracy. Extensive experiments on different MRI datasets demonstrate that our MSM-Seg framework outperforms state-of-the-art methods in multi-modal metastases and glioma tumor segmentation. The code is available at https://github.com/xq141839/MSM-Seg.

cs.CV↗

arXiv2025

LLM-JEPA: Large Language Models Meet Joint Embedding Predictive Architectures

Large Language Model (LLM) pretraining, finetuning, and evaluation rely on input-space reconstruction and generative capabilities. Yet, it has been observed in vision that embedding-space training objectives, e.g., with Joint Embedding Predictive Architectures (JEPAs), are far superior to their input-space counterpart. That mismatch in how training is achieved between language and vision opens up a natural question: {\em can language training methods learn a few tricks from the vision ones?} The lack of JEPA-style LLM is a testimony of the challenge in designing such objectives for language. In this work, we propose a first step in that direction where we develop LLM-JEPA, a JEPA based solution for LLMs applicable both to finetuning and pretraining. Thus far, LLM-JEPA is able to outperform the standard LLM training objectives by a significant margin across models, all while being robust to overfiting. Those findings are observed across numerous datasets (NL-RX, GSM8K, Spider, RottenTomatoes) and various models from the Llama3, OpenELM, Gemma2 and Olmo families. Code: https://github.com/rbalestr-lab/llm-jepa.

cs.CL↗

arXiv2025

LoRA Users Beware: A Few Spurious Tokens Can Manipulate Your Finetuned Model

Large Language Models (LLMs) are commonly finetuned for a variety of use cases and domains. A common approach is to leverage Low-Rank Adaptation (LoRA) -- known to provide strong performance at low resource costs. In this study, we demonstrate that LoRA actually opens the door to short-cut vulnerabilities -- and the more resource efficient is the LoRA setup, the more vulnerable will be the finetuned model to aggressive attacks. To measure that vulnerability, we introduce Seamless Spurious Token Injection (SSTI), where we find that LoRA exclusively focuses on even just a single token that is spuriously correlated with downstream labels. In short, injection of that spurious token during finetuning ensure that the model's prediction at test-time can be manipulated on-demand. We conducted experiments across model families and datasets to evaluate the impact of SSTI during LoRA finetuning while providing possible mitigations. Our experiments conclude that none of the existing checkers and preprocessors can sanitize a dataset raising new concerns for data quality and AI safety.

cs.LG↗

arXiv2025

Directed Information $γ$-covering: An Information-Theoretic Framework for Context Engineering

We introduce \textbf{Directed Information $γ$-covering}, a simple but general framework for redundancy-aware context engineering. Directed information (DI), a causal analogue of mutual information, measures asymmetric predictiveness between chunks. If $\operatorname{DI}_{i \to j} \ge H(C_j) - γ$, then $C_i$ suffices to represent $C_j$ up to $γ$ bits. Building on this criterion, we formulate context selection as a $γ$-cover problem and propose a greedy algorithm with provable guarantees: it preserves query information within bounded slack, inherits $(1+\ln n)$ and $(1-1/e)$ approximations from submodular set cover, and enforces a diversity margin. Importantly, building the $γ$-cover is \emph{query-agnostic}: it incurs no online cost and can be computed once offline and amortized across all queries. Experiments on HotpotQA show that $γ$-covering consistently improves over BM25, a competitive baseline, and provides clear advantages in hard-decision regimes such as context compression and single-slot prompt selection. These results establish DI $γ$-covering as a principled, self-organizing backbone for modern LLM pipelines.

cs.IT↗

arXiv2025

RecBase: Generative Foundation Model Pretraining for Zero-Shot Recommendation

Recent advances in LLM-based recommendation have shown promise, yet their cross-domain generalization is hindered by a fundamental mismatch between language-centric pretraining and the recommendation task. Existing methods, relying on language-level knowledge, fail to capture dynamic, item-level user interests across domains. To bridge this gap, we propose RecBase, a domain-agnostic foundational model pretrained with a recommendation-oriented objective. RecBase leverages a large-scale, heterogeneous, cross-domain corpus with unified textual representations and feature mappings to enhance cross-domain generalization. To further align item semantics across domains, we introduce a unified item tokenizer that encodes items into hierarchical concept identifiers, enabling structured representation and efficient vocabulary sharing. The model is trained using an autoregressive objective to capture complex item-level sequential patterns. On eight real-world datasets, our 1.5B-parameter model matches or surpasses the performance of LLM baselines up to 7B parameters in zero-shot and cross-domain recommendation tasks.

cs.IR↗

arXiv2025

HydroGAT: Distributed Heterogeneous Graph Attention Transformer for Spatiotemporal Flood Prediction

Accurate flood forecasting remains a challenge for water-resource management, as it demands modeling of local, time-varying runoff drivers (e.g., rainfall-induced peaks, baseflow trends) and complex spatial interactions across a river network. Traditional data-driven approaches, such as convolutional networks and sequence-based models, ignore topological information about the region. Graph Neural Networks (GNNs) propagate information exactly along the river network, which is ideal for learning hydrological routing. However, state-of-the-art GNN-based flood prediction models collapse pixels to coarse catchment polygons as the cost of training explodes with graph size and higher resolution. Furthermore, most existing methods treat spatial and temporal dependencies separately, either applying GNNs solely on spatial graphs or transformers purely on temporal sequences, thus failing to simultaneously capture spatiotemporal interactions critical for accurate flood prediction. We introduce a heterogenous basin graph where every land and river pixel is a node connected by physical hydrological flow directions and inter-catchment relationships. We propose HydroGAT, a spatiotemporal network that adaptively learns local temporal importance and the most influential upstream locations. Evaluated in two Midwestern US basins and across five baseline architectures, our model achieves higher NSE (up to 0.97), improved KGE (up to 0.96), and low bias (PBIAS within $\pm$5%) in hourly discharge prediction, while offering interpretable attention maps that reveal sparse, structured intercatchment influences. To support high-resolution basin-scale training, we develop a distributed data-parallel pipeline that scales efficiently up to 64 NVIDIA A100 GPUs on NERSC Perlmutter supercomputer, demonstrating up to 15x speedup across machines. Our code is available at https://github.com/swapp-lab/HydroGAT.

cs.LG↗

arXiv2025

X-PRINT:Platform-Agnostic and Scalable Fine-Grained Encrypted Traffic Fingerprinting

Although encryption protocols such as TLS are widely de-ployed,side-channel metadata in encrypted traffic still reveals patterns that allow application and behavior inference.How-ever,existing fine-grained fingerprinting approaches face two key limitations:(i)reliance on platform-dependent charac-teristics,which restricts generalization across heterogeneous platforms,and(ii)poor scalability for fine-grained behavior identification in open-world settings. In this paper,we present X-PRINT,the first server-centric,URI-based framework for cross-platform fine-grained encrypted-traffic fingerprinting.X-PRINT systematically demonstrates that backend URI invocation patterns can serve as platform-agnostic invariants and are effective for mod-eling fine-grained behaviors.To achieve robust identifica-tion,X-PRINT further leverages temporally structured URI maps for behavior inference and emphasizes the exclusion of platform-or application-specific private URIs to handle unseen cases,thereby improving reliability in open-world and cross-platform settings.Extensive experiments across diverse cross-platform and open-world settings show that X-PRINT achieves state-of-the-art accuracy in fine-grained fingerprint-ing and exhibits strong scalability and robustness.

cs.CR↗