arXiv subjects

Lei Gao

Publications and source records attributed to Lei Gao.

At least 37 records · Page 2Linked to original sources

arXiv2026

An Ultra-Low Power and Fast Ising Machine using Voltage-Controlled Magnetoresistive Random Access Memory

Physics-inspired computing paradigms, such as Ising machines, are emerging as promising hardware alternatives to traditional von Neumann architectures for tackling computationally intensive combinatorial optimization problems (COPs). While quantum, optical, and electronic devices have garnered significant attention for their potential in realizing Ising machines, their translation into practical systems for industry-relevant applications remains challenging, with each approach facing specific limitations in power consumption and speed. To address this challenge, we report the first chip-level spintronic Ising machine using voltage-controlled magnetoresistive random access memory. The core of our design leverages magnetic tunnel junctions (MTJs) driven by the voltage-controlled magnetic anisotropy effect to realize the probabilistic update of Ising spins through a new mechanism. It enables a latency below 1 ns and an energy consumption under 40 fJ per spin update, achieving a 1000-times improvement over previous current-driven MTJ-based implementations. We map two real-world COPs in electronic design automation-global routing and layer assignment-onto the Ising model and demonstrate high-quality results with an energy efficiency of 25000 solutions per second per watt. This outperforms state-of-the-art quantum and graphics processing units by six and seven orders of magnitude, respectively. These results establish voltage-controlled spintronics as a compelling route towards next-generation physics-inspired machine intelligence, offering a paradigm for ultra-low-power, high-speed, and scalable computation.

physics.app-ph↗

arXiv2026

Dynamic Dispersion Accumulation in Fiber Loops for Realizing Record-High Frequency Resolution or Ultra-Low Signal Sampling Rate in Dispersion-Based Photonics-Assisted Wideband Microwave Measurement Systems

Dispersion-based photonics-assisted microwave measurement systems provide immense potential for real-time analysis of wideband and dynamic signals. However, they face two critical challenges: a difficulty in achieving high frequency resolution over a wideband analysis bandwidth, and a reliance on large-bandwidth-and-high-sampling-rate oscilloscopes to capture the resulting ultra-narrow pulses. We introduce a dynamic dispersion accumulation technique to overcome these limitations. By circulating the optical signal in fiber loops containing a dispersion-compensating fiber, we achieve a high accumulated dispersion of -215700 ps/nm. This high dispersion relaxes the required chirp rate of the chirped optical signal, enabling two distinct advantages: When the analysis bandwidth is fixed, a lower chirp rate enables a longer temporal period, yielding a record-high frequency resolution of 27.9 MHz; When the temporal period is fixed, a lower chirp rate enables a smaller bandwidth, generating a wider pulse and thus relaxing pulse sampling requirements at the expense of analysis bandwidth. This sacrifice in analysis bandwidth can be compensated by a duty-cycle-enabling technique, which holds the potential to extend the analysis bandwidth beyond 100 GHz. This work breaks the performance and hardware limitations in dispersion-based systems, paving the way for high frequency resolution, wideband microwave measurement systems that are both real-time and cost-effective.

physics.optics↗

arXiv2026

Broadband tunable narrow-linewidth laser based on scattering-enhanced fiber covering E-S-C-L bands

This work demonstrates a broadband tunable narrow-linewidth laser based on scattering-enhanced fiber, covering the E-S-C-L wavelength bands from 1337.47 nm to 1631.39 nm, with a total tuning span of 293.92 nm. The laser employs two semiconductor optical amplifiers (SOAs) centered at 1420 nm and 1550 nm, which are connected into a single ring resonator via polarization multiplexing. Wavelength selection and tunability is realized using an ultra-broadband tunable filter based on a blazed grating. To suppress side longitude modes, an 18-meter-long femtosecond-laser-empowered random scattering fiber is utilized inside the cavity as a feedback medium, yielding an output linewidths between 1.54 kHz and 2.61 kHz. Benefited from the fast response of the galvanometer mirror and short relaxation time of SOAs, wavelength switching time is less than 1 ms under different tuning channels among the wavelength range of near 300 nm. The stable single-longitude-mode operation is maintained across the entire tuning range. The exceptionally broad tuning range and high spectral purity of the laser endow it with significant application potentials across a wide range of fields.

physics.optics↗

arXiv2025

Broadband tunable microwave photonic radar for simultaneous detection of human respiration, heartbeat, and speech with deep learning-based speech recognition

Multimodal vital sign monitoring and speech detection hold significant importance in medical health, public safety, and other fields. This study proposes a broadband tunable microwave photonic radar system that can simultaneously monitor respiration, heartbeat, and speech. The system works by generating broadband radar signals to detect subtle skin displacements caused by these physiological activities. It then utilizes phase variations in radar echo signals to extract and reconstruct the corresponding physiological signals. In order to enhance the processing capability for speech signals, a convolutional neural network with a dual-channel feature fusion model is incorporated, enabling high-precision speech recognition. In addition, the system's frequency-tunable characteristic allows it to flexibly switch frequency bands to adapt to different working environments, greatly improving its practicality and environmental adaptability. In concept-verification experiments, speech signals were reconstructed and recognized in the Ku, K, and Ka bands, achieving recognition accuracies of 97.20%, 98.07%, and 97.43%, respectively. The system's capability to detect multimodal vital signs was also thoroughly validated using a respiratory and heartbeat simulator. During a 20-second monitoring period, while accurately reconstructing speech, the maximum average error counts for respiratory and heartbeat monitoring were 0.39 and 0.87, respectively, proving its reliability and effectiveness in multimodal vital sign monitoring.

physics.optics↗

arXiv2025

Itinerant Orbital Hall Effect Mechanism Leading to Large Negative Orbital Torques from Light Metal Vanadium

The orbital Hall effect (OHE) has attracted significant attention for developing energy-efficient electronic devices. However, utilizing it in fast, low-power devices requires an enhanced understanding of underlying extrinsic and intrinsic contributions to OHE at timescales ranging from quasi-static to picoseconds. Here, we investigate OHE in light metal vanadium (V) using a combination of selected measurement schemes, spanning the full frequency range. We observe a negative damping-like torque efficiency from V, opposite to conventional theoretical predictions, with a magnitude that depends on the adjacent ferromagnet, a dependence that indicates orbital effects. These results, with consistent torque efficiencies across all frequencies, corroborate a negative and intrinsic OHE in V with a large effective orbital Hall conductivity of $-(1.44 \pm 0.34)\,\frac{\hbar}{2e}\,\times 10^{5}\,Ω^{-1}\,\mathrm{m}^{-1}$ and a long orbital diffusion length of $(15.0 \pm 2.5)\,\mathrm{nm}$. To explain the observed OHE, we develop a theoretical model incorporating both local and itinerant circulation contributions to OHE. The model agrees excellently with the experimental results, demonstrating that itinerant contributions are essential for a complete physical understanding of intrinsic OHE. Our consistent experimental and theoretical data highlight the importance of itinerant contributions governing the fundamental understanding of intrinsic OHE and the large effects found open pathways for energy-efficient orbitronic devices.

cond-mat.mes-hall↗

arXiv2025

Striking the Right Balance between Compute and Copy: Improving LLM Inferencing Under Speculative Decoding

With the skyrocketing costs of GPUs and their virtual instances in the cloud, there is a significant desire to use CPUs for large language model (LLM) inference. KV cache update, often implemented as allocation, copying, and in-place strided update for each generated token, incurs significant overhead. As the sequence length increases, the allocation and copy overheads dominate the performance. Alternate approaches may allocate large KV tensors upfront to enable in-place updates, but these matrices (with zero-padded rows) cause redundant computations. In this work, we propose a new KV cache allocation mechanism called Balancing Memory and Compute (BMC). BMC allocates, once every r iterations, KV tensors with r redundant rows, allowing in-place update without copy overhead for those iterations, but at the expense of a small amount of redundant computation. Second, we make an interesting observation that the extra rows allocated in the KV tensors and the resulting redundant computation can be repurposed for Speculative Decoding (SD) that improves token generation efficiency. Last, BMC represents a spectrum of design points with different values of r. To identify the best-performing design point(s), we derive a simple analytical model for BMC. The proposed BMC method achieves an average throughput acceleration of up to 3.2x over baseline HuggingFace (without SD). Importantly when we apply BMC with SD, it results in an additional speedup of up to 1.39x, over and above the speedup offered by SD. Further, BMC achieves a throughput acceleration of up to 1.36x and 2.29x over state-of-the-art inference servers vLLM and DeepSpeed, respectively. Although the BMC technique is evaluated extensively across different classes of CPUs (desktop and server class), we also evaluate the scheme with GPUs and demonstrate that it works well for GPUs.

cs.DC↗

arXiv2025

ViFP: A Framework for Visual False Positive Detection to Enhance Reasoning Reliability in VLMs

During reasoning in vision-language models (VLMs), false positive (FP) reasoning occurs when a model produces the correct answer but follows an incorrect reasoning path, resulting in undermined reasoning reliability. Existing approaches mainly rely on prompt engineering, knowledge distillation or reinforcement learning to improve reasoning reliability, both of which require large amounts of high-quality data and thus limit practical applicability. Few approaches have focused on directly detecting and correcting FPs. To address these issues, we propose ViFP, a framework for Visual False Positive Detection to Enhance Reasoning Reliability in VLMs. ViFP builds effective reasoning paths through multi-turn QA and dynamically analyzes the consistency of the reasoning path to identify potential FPs. It also introduces a targeted reasoning chain correction mechanism to modify FP reasoning, thereby improving logical consistency and accuracy. Finally, we introduce a reliability evaluation metric, VoC, which integrates answer accuracy and the FP rate, providing a quantitative tool to assess whether a VLM not only answers correctly but also reasons reliably. Our experiments on closed-source VLMs show that ViFP consistently improves performance across three datasets: A-OKVQA, OK-VQA, and FVQA. On A-OKVQA, ViFP improves accuracy by up to 5.4%, surpassing the previous state-of-the-art by 4.3%, and significantly reduces the number of FPs, validating its benefits in enhancing reasoning reliability.

cs.CV↗

arXiv2025

Achieving Empirical Potential Efficiency with DFT Accuracy: A Neuroevolution Potential for the $α$-Fe--C--H System

A neuroevolution potential (NEP) for the ternary $α$-Fe--C--H system was developed based on a database generated from spin-polarized density functional theory (DFT) calculations, achieving empirical potential efficiency with DFT accuracy. At the same power consumption, simulation speeds using NEP are comparable to, or even faster than, those with bond order potentials. The NEP achieves DFT-level accuracy across a wide range of scenarios commonly encountered in studies of $α$-Fe- and $α$-Fe--C under hydrogen environments. The NEP enables large-scale atomistic simulations with DFT-level accuracy at the cost of empirical potentials, offering a practical tool to study hydrogen embrittlement in steel.

cond-mat.mtrl-sci↗

arXiv2025

DistilLock: Safeguarding LLMs from Unauthorized Knowledge Distillation on the Edge

Large Language Models (LLMs) have demonstrated strong performance across diverse tasks, but fine-tuning them typically relies on cloud-based, centralized infrastructures. This requires data owners to upload potentially sensitive data to external servers, raising serious privacy concerns. An alternative approach is to fine-tune LLMs directly on edge devices using local data; however, this introduces a new challenge: the model owner must transfer proprietary models to the edge, which risks intellectual property (IP) leakage. To address this dilemma, we propose DistilLock, a TEE-assisted fine-tuning framework that enables privacy-preserving knowledge distillation on the edge. In DistilLock, a proprietary foundation model is executed within a trusted execution environment (TEE) enclave on the data owner's device, acting as a secure black-box teacher. This setup preserves both data privacy and model IP by preventing direct access to model internals. Furthermore, DistilLock employs a model obfuscation mechanism to offload obfuscated weights to untrusted accelerators for efficient knowledge distillation without compromising security. We demonstrate that DistilLock prevents unauthorized knowledge distillation processes and model-stealing attacks while maintaining high computational efficiency, but offering a secure and practical solution for edge-based LLM personalization.

cs.CR↗

arXiv2025

Ignition Phase : Standard Training for Fast Adversarial Robustness

Adversarial Training (AT) is a cornerstone defense, but many variants overlook foundational feature representations by primarily focusing on stronger attack generation. We introduce Adversarial Evolution Training (AET), a simple yet powerful framework that strategically prepends an Empirical Risk Minimization (ERM) phase to conventional AT. We hypothesize this initial ERM phase cultivates a favorable feature manifold, enabling more efficient and effective robustness acquisition. Empirically, AET achieves comparable or superior robustness more rapidly, improves clean accuracy, and cuts training costs by 8-25\%. Its effectiveness is shown across multiple datasets, architectures, and when augmenting established AT methods. Our findings underscore the impact of feature pre-conditioning via standard training for developing more efficient, principled robust defenses. Code is available in the supplementary material.

cs.LG↗

arXiv2025

Nitrogen-Triggered Amorphization Enables High-Performance Solid-State Electrolytes

Amorphous solid-state electrolytes (SSEs) hold great promise for advancing the application of all-solid-state batteries (ASSBs), owing to their favorable ionic conductivity, structural tunability, and promising electrochemical performance. However, the absence of universal design principles for amorphous SSEs limits their development. By fundamentally re-evaluating the amorphization-forming ability of amorphous SSE systems, this study establishes a nitrogen-driven universal strategy to convert diverse metal chlorides into amorphous xLi3N-MCly (0.3 < 3x < 1.9; M denotes a metal element; 2 < y < 5) SSE. Nitrogen synergistically disrupts crystalline order via distorted coordination polyhedra and N-bridged networks, while dynamic bond reorganization enables rapid Li+ migration, achieving ionic conductivity of 2.02 mS cm-1 for 0.533Li3N-HfCl4 at 25 °C. Structural-property relationships reveal that high charge density and bridging capability of N3- enhance network disorder, shorten metal coordinating atom distances, and optimize Li+ diffusion pathway connectivity. ASSBs employing 0.533Li3N-HfCl4 retain 81.87% capacity after 2000 cycles at 1000 mA g-1 with high cathode loading (6.24 mg cm-2), demonstrating engineering viability. This work provides a paradigm for rational design of high-performance amorphous SSEs.

cond-mat.mtrl-sci↗

arXiv2025

MobiZO: Enabling Efficient LLM Fine-Tuning at the Edge via Inference Engines

Large Language Models (LLMs) are currently pre-trained and fine-tuned on large cloud servers. The next frontier is LLM personalization, where a foundation model can be fine-tuned with user/task-specific data. Given the sensitive nature of such private data, it is desirable to fine-tune these models on edge devices to improve user trust. However, fine-tuning on resource-constrained edge devices presents significant challenges due to substantial memory and computational demands, as well as limited infrastructure support. We observe that inference engines (e.g., ExecuTorch) can be repurposed for fine-tuning by leveraging zeroth-order (ZO) optimization, which uses multiple forward passes to approximate gradients. While promising, direct application of ZO methods on edge devices is inefficient due to the high computational cost of multiple forward passes required for accurate gradient estimation, and their deployment has been largely unexplored in practice. We introduce MobiZO, a resource-efficient fine-tuning framework for LLMs specifically designed for edge devices. MobiZO combines three key innovations: (1) a parallelized randomized gradient estimator that employs both outer-loop and inner-loop parallelism to eliminate sequential forward passes, (2) a specialized Multi-Perturbed LoRA (MP-LoRA) module that enables efficient realization of both inner and outer loop parallelism, and (3) a seamless integration with ExecuTorch for on-device training, requiring no modifications to the runtime. Experiments demonstrate that MobiZO achieves substantial runtime speedups and memory savings while improving fine-tuning accuracy, paving the way for practical deployment of LLMs in real-time, on-device applications.

cs.LG↗

arXiv2025

Direct observation of photonic spin Hall effect in Mie scattering

The photonic spin Hall effect (PSHE), a hallmark of spin-orbit interaction of light, has long been considered a promising route toward spin-controlled functionalities in nanophotonics. Yet, its practical realization has been severely limited by the inherently weak spin-orbit coupling in typical systems, resulting in vanishingly small transverse shifts and extremely low scattering efficiency. This fundamental trade-off has rendered the PSHE observable only through complex weak measurement protocols and signal amplification-approaches that come at the cost of further intensity loss, particularly in nanoscale systems. In this work, we overcome this longstanding challenge by introducing a novel mechanism based on symmetry breaking and mode coupling in a standalone scatterer, which unlocks a regime of Friedrich-Wintgen superscattering with strong near-field spin-orbit interaction. This allows for simultaneous enhancement of both the photonic spin Hall shift and the far-field scattering intensity-boosting the latter by nearly two orders of magnitude compared to conventional dipolar particles. Through tailored multipolar interference, the PSHE is made accessible at experimentally convenient angles, enabling post selection-free detection. We report the first direct experimental observation of the PSHE from a single superscattering particle, achieved in the microwave regime via polarization-resolved far-field measurements. Our findings not only validate a new physical pathway for enhancing spin-dependent light-matter interactions, but also establish a robust, scalable platform for spin-based photonic technologies. This breakthrough opens new avenues in precision optical metrology, advanced imaging, LIDAR systems, and integrated photonic circuitry, bridging a critical gap between fundamental spin optics and real-world applications.

physics.optics↗

arXiv2025

DEL: Context-Aware Dynamic Exit Layer for Efficient Self-Speculative Decoding

Speculative Decoding (SD) is a widely used approach to accelerate the inference of large language models (LLMs) without reducing generation quality. It operates by first using a compact model to draft multiple tokens efficiently, followed by parallel verification using the target LLM. This approach leads to faster inference compared to auto-regressive decoding. While there are multiple approaches to create a draft model, one promising approach is to use early-exit methods. These methods draft candidate tokens by using a subset of layers of the primary model and applying the remaining layers for verification, allowing a single model to handle both drafting and verification. While this technique reduces memory usage and computational cost, its performance relies on the choice of the exit layer for drafting and the number of tokens drafted (speculation length) in each SD round. Prior works use hyperparameter exploration to statically select these values. However, our evaluations show that these hyperparameter values are task-specific, and even within a task they are dependent on the current sequence context. We introduce DEL (Dynamic Exit Layer), a plug-and-play method that adaptively selects the exit layer and speculation length during inference. DEL dynamically tracks the token acceptance rate if the tokens are drafted at each layer of an LLM and uses that knowledge to heuristically select the optimal exit layer and speculation length. Our experiments across a broad range of models and downstream tasks show that DEL achieves overall speedups of $2.16\times$$\sim$$2.62\times$ over vanilla auto-regressive decoding and improves upon state-of-the-art SD methods, which peak at $2.43\times$, by up to $0.19\times$. The code is available at https://github.com/hoenza/DEL.

cs.CL↗

arXiv2025

Switchable Exchange Bias Resulting from Correlated Domain Structures in Orthogonally Coupled Antiferromagnet/Ferromagnet van der Waals Heterostructures

Van der Waals (vdW) magnetic heterostructures offer a versatile platform for engineering interfacial spin interactions with atomic precision, enabling nontrivial spin textures and dynamic behaviors. In this work, we report robust asymmetric magnetization reversal and exchange bias in Fe3GeTe2 (FGT), driven by interlayer exchange coupling with the A-type antiferromagnet CrSBr. Despite the orthogonal magnetic anisotropies out-of-plane easy axis in FGT and in-plane in CrSBr, we observe a strong interfacial exchange interaction that gives rise to pronounced and switchable exchange bias and asymmetric switching in FGT, persisting up to the Néel temperature of CrSBr (132 K) as revealed by anomalous Hall effect measurements. We uncover the microscopic origin of this behavior through cross-sectional magnetic imaging of the domain structure using off-axis electron holography. The results reveal that the asymmetric switching and exchange bias arise from the influence of CrSBr on the domain configuration of FGT, where the in-plane antiferromagnetic state of CrSBr promotes the formation of stripe-like domain structures in FGT with circular rotation of magnetization in the cross-sectional bc plane defined by the easy axes of both FGT and CrSBr. These findings elucidate the mechanism of exchange bias in orthogonally coupled van der Waals systems and demonstrate a pathway for stabilizing three-dimensional domain structures in ferromagnets through interfacial exchange interactions.

cond-mat.mtrl-sci↗

arXiv2025

HAMLET-FFD: Hierarchical Adaptive Multi-modal Learning Embeddings Transformation for Face Forgery Detection

The rapid evolution of face manipulation techniques poses a critical challenge for face forgery detection: cross-domain generalization. Conventional methods, which rely on simple classification objectives, often fail to learn domain-invariant representations. We propose HAMLET-FFD, a cognitively inspired Hierarchical Adaptive Multi-modal Learning framework that tackles this challenge via bidirectional cross-modal reasoning. Building on contrastive vision-language models such as CLIP, HAMLET-FFD introduces a knowledge refinement loop that iteratively assesses authenticity by integrating visual evidence with conceptual cues, emulating expert forensic analysis. A key innovation is a bidirectional fusion mechanism in which textual authenticity embeddings guide the aggregation of hierarchical visual features, while modulated visual features refine text embeddings to generate image-adaptive prompts. This closed-loop process progressively aligns visual observations with semantic priors to enhance authenticity assessment. By design, HAMLET-FFD freezes all pretrained parameters, serving as an external plugin that preserves CLIP's original capabilities. Extensive experiments demonstrate its superior generalization to unseen manipulations across multiple benchmarks, and visual analyses reveal a division of labor among embeddings, with distinct representations specializing in fine-grained artifact recognition.

cs.CV↗

arXiv2025

KVPR: Efficient LLM Inference with I/O-Aware KV Cache Partial Recomputation

Inference for Large Language Models (LLMs) is computationally demanding. To reduce the cost of auto-regressive decoding, Key-Value (KV) cache is used to store intermediate activations, which significantly lowers the computational overhead for token generation. However, the memory required for the KV cache grows rapidly, often exceeding the capacity of GPU memory. A cost-effective alternative is to offload KV cache to CPU memory, which alleviates GPU memory pressure, but shifts the bottleneck to the limited bandwidth of the PCIe connection between the CPU and GPU. Existing methods attempt to address these issues by overlapping GPU computation with I/O or employing CPU-GPU heterogeneous execution, but they are hindered by excessive data movement and dependence on CPU capabilities. Fully overlapping PCIe communication latency gets challenging as the size of the KV cache grows and/or the GPU compute capabilities increase. In this paper, we introduce KVPR, an efficient I/O-aware LLM inference method where the CPU first transfers a partial set of activations, from which the GPU can start recomputing the KV cache values. While the GPU recomputes the partial KV cache, the remaining portion of the KV cache is transferred concurrently from the CPU. This approach overlaps GPU recomputation with KV cache transfer to minimize idle GPU time and maximize inference performance. KVPR is fully automated by integrating a profiler module that utilizes input characteristics and system hardware information, a scheduler module to optimize the distribution of computation and communication workloads, and a runtime module to efficiently execute the derived execution plan. Experimental results show that KVPR achieves up to 35.8% lower latency and 46.2% higher throughput during decoding compared to state-of-the-art approaches. The code is available at https://github.com/chaoyij/KVPR.

cs.LG↗

arXiv2025

Advanced microwave photonic waveform editing: enabling the evolution of radar systems into joint radar and spectrum sensing systems

In response to the urgent demand for the development of future radar application platforms from single radar functionality towards integrated multi-functional systems, we show an advanced microwave photonic waveform editing method that enables the editing of arbitrary radar waveforms, equipping them with the capability to perform spectrum sensing. This, in turn, expands single-function radar systems into joint radar and spectrum sensing systems. We theoretically define and calculate the accumulation function of an arbitrary waveform after passing through a specific dispersive medium, and utilize this accumulation function to further design a corresponding binary sequence for editing the waveform. After editing, the accumulation function of the edited waveform approximates that of a linearly frequency-modulated signal matching the specific dispersive medium. Thus, the edited waveform can be compressed into a narrow pulse after passing through the dispersive medium, realizing the frequency-to-time mapping for achieving frequency measurement or time-frequency analysis. The concept is verified by a simulation and an experiment. Using a dispersion compensating fiber with a total dispersion of -6817 ps/nm, arbitrary waveforms, including a 7-bit Barker phase-coded waveform, a linearly frequency-modulated waveform, a nonlinearly frequency-modulated waveform, and a waveform with an "E" time-frequency diagram, are edited and further used for microwave frequency measurement and time-frequency analysis in an ultra-wide bandwidth of 36.8 GHz. The temporal resolution and frequency resolution are 2 ns and 0.86 GHz, respectively.

physics.optics↗