arXiv subjects

Donghyun Lee

Publications and source records attributed to Donghyun Lee.

At least 37 records · Page 2Linked to original sources

arXiv2026

TT-SNN: Tensor Train Decomposition for Efficient Spiking Neural Network Training

Spiking Neural Networks (SNNs) have gained significant attention as a potentially energy-efficient alternative for standard neural networks with their sparse binary activation. However, SNNs suffer from memory and computation overhead due to spatio-temporal dynamics and multiple backpropagation computations across timesteps during training. To address this issue, we introduce Tensor Train Decomposition for Spiking Neural Networks (TT-SNN), a method that reduces model size through trainable weight decomposition, resulting in reduced storage, FLOPs, and latency. In addition, we propose a parallel computation pipeline as an alternative to the typical sequential tensor computation, which can be flexibly integrated into various existing SNN architectures. To the best of our knowledge, this is the first of its kind application of tensor decomposition in SNNs. We validate our method using both static and dynamic datasets, CIFAR10/100 and N-Caltech101, respectively. We also propose a TT-SNN-tailored training accelerator to fully harness the parallelism in TT-SNN. Our results demonstrate substantial reductions in parameter size (7.98X), FLOPs (9.25X), training time (17.7%), and training energy (28.3%) during training for the N-Caltech101 dataset, with negligible accuracy degradation.

cs.NE↗

arXiv2026

Ill-posedness of the Boltzmann-BGK model in the exponential class

BGK (Bhatnagar-Gross-Krook) model is a relaxation-type model of the Boltzmann equation, which is popularly used in place of the Boltzmann equation in physics and engineering. In this paper, we address the ill-posedness problem for the BGK model, in which the solution instantly escapes the initial solution space. For this, we propose two ill-posedness scenarios, namely, the homogeneous and the inhomogeneous ill-posedness mechanisms. In the former case, we find a class of spatially homogeneous solutions to the BGK model, where removing the small velocity part of the initial data triggers ill-posedness by increasing temperature. For the latter, we construct a spatially inhomogeneous solution to the BGK model such that the local temperature constructed from the solution has a polynomial growth in spatial variable. These ill-posedness properties for the BGK model pose a stark contrast with the Boltzmann equation for which the solution map is, at least for a finite time, stable in the corresponding solution spaces.

math.AP↗

arXiv2026

Building a Mind Palace: Structuring Environment-Grounded Semantic Graphs for Effective Long Video Analysis with LLMs

Long-form video understanding with Large Vision Language Models is challenged by the need to analyze temporally dispersed yet spatially concentrated key moments within limited context windows. In this work, we introduce VideoMindPalace, a new framework inspired by the "Mind Palace", which organizes critical video moments into a topologically structured semantic graph. VideoMindPalace organizes key information through (i) hand-object tracking and interaction, (ii) clustered activity zones representing specific areas of recurring activities, and (iii) environment layout mapping, allowing natural language parsing by LLMs to provide grounded insights on spatio-temporal and 3D context. In addition, we propose the Video MindPalace Benchmark (VMB), to assess human-like reasoning, including spatial localization, temporal reasoning, and layout-aware sequential understanding. Evaluated on VMB and established video QA datasets, including EgoSchema, NExT-QA, IntentQA, and the Active Memories Benchmark, VideoMindPalace demonstrates notable gains in spatio-temporal coherence and human-aligned reasoning, advancing long-form video analysis capabilities in VLMs.

cs.CV↗

arXiv2026

Memba: Membrane-driven Parameter-Efficient Fine-Tuning for Mamba

State Space Models (SSMs) have emerged as powerful alternatives to attention-based Transformers, with Mamba demonstrating impressive efficiency and scalability. As these models grow increasingly larger, the need for Parameter-Efficient Fine-Tuning (PEFT) methods becomes critical to adapt pre-trained Mamba to downstream tasks without prohibitive computational costs. However, previous approaches simply apply traditional Transformer-tailored PEFT methods without addressing the unique temporal processing dynamics of SSMs. To address this limitation, we propose Memba, a membrane-driven PEFT approach specifically designed for Mamba. Memba introduces Leaky Integrate Membrane (LIM) neurons as bio-inspired gating mechanisms that naturally accumulate membrane potentials over time, enhancing selective information retention. By strategically combining LIM neurons with Low-Rank Adaptations (LoRA) and cross-layer membrane transfer, our approach significantly improves Mamba's temporal modeling capabilities. Extensive experiments across language and vision tasks demonstrate that Memba achieves substantial improvements over existing PEFT methods. The code is available at https://github.com/Intelligent-Computing-Lab-Yale/Memba.

cs.LG↗

arXiv2026

OpenWorldSAM: Extending SAM2 for Universal Image Segmentation with Language Prompts

The ability to segment objects based on open-ended language prompts remains a critical challenge, requiring models to ground textual semantics into precise spatial masks while handling diverse and unseen categories. We present OpenWorldSAM, a framework that extends the prompt-driven Segment Anything Model v2 (SAM2) to open-vocabulary scenarios by integrating multi-modal embeddings extracted from a lightweight vision-language model (VLM). Our approach is guided by four key principles: i) Unified prompting: OpenWorldSAM supports a diverse range of prompts, including category-level and sentence-level language descriptions, providing a flexible interface for various segmentation tasks. ii) Efficiency: By freezing the pre-trained components of SAM2 and the VLM, we train only 4.5 million parameters on the COCO-stuff dataset, achieving remarkable resource efficiency. iii) Instance Awareness: We enhance the model's spatial understanding through novel positional tie-breaker embeddings and cross-attention layers, enabling effective segmentation of multiple instances. iv) Generalization: OpenWorldSAM exhibits strong zero-shot capabilities, generalizing well on unseen categories and an open vocabulary of concepts without additional training. Extensive experiments demonstrate that OpenWorldSAM achieves state-of-the-art performance in open-vocabulary semantic, instance, and panoptic segmentation across multiple benchmarks. Code is available at https://github.com/GinnyXiao/OpenWorldSAM.

cs.CV↗

arXiv2026

MD-SNN: Membrane Potential-aware Distillation on Quantized Spiking Neural Network

Spiking Neural Networks (SNNs) offer a promising and energy-efficient alternative to conventional neural networks, thanks to their sparse binary activation. However, they face challenges regarding memory and computation overhead due to complex spatio-temporal dynamics and the necessity for multiple backpropagation computations across timesteps during training. To mitigate this overhead, compression techniques such as quantization are applied to SNNs. Yet, naively applying quantization to SNNs introduces a mismatch in membrane potential, a crucial factor for the firing of spikes, resulting in accuracy degradation. In this paper, we introduce Membrane-aware Distillation on quantized Spiking Neural Network (MD-SNN), which leverages membrane potential to mitigate discrepancies after weight, membrane potential, and batch normalization quantization. To our knowledge, this study represents the first application of membrane potential knowledge distillation in SNNs. We validate our approach on various datasets, including CIFAR10, CIFAR100, N-Caltech101, and TinyImageNet, demonstrating its effectiveness for both static and dynamic data scenarios. Furthermore, for hardware efficiency, we evaluate the MD-SNN with SpikeSim platform, finding that MD-SNNs achieve 14.85X lower energy-delay-area product (EDAP), 2.64X higher TOPS/W, and 6.19X higher TOPS/mm2 compared to floating point SNNs at iso-accuracy on N-Caltech101 dataset.

cs.NE↗

arXiv2025

DuoGPT: Training-free Dual Sparsity through Activation-aware Pruning in LLMs

Large language models (LLMs) deliver strong performance but are difficult to deploy due to high memory and compute costs. While pruning reduces these demands, most methods ignore activation sparsity observed at runtime. We reinterpret activation sparsity as dynamic structured weight sparsity and propose DuoGPT, a unified framework that constructs dual-sparse (spMspV) workloads by combining unstructured weight pruning with activation sparsity. To preserve accuracy, we extend the Optimal Brain Compression (OBC) framework with activation-aware calibration and introduce output residuals from the dense model as correction terms. We further optimize the solution for efficient GPU execution, enabling scalability to billion-parameter LLMs. Evaluations on LLaMA-2 and LLaMA-3 show that DuoGPT outperforms state-of-the-art structured pruning methods by up to 9.17% accuracy at an iso-speedup of 1.39$\times$ compared to the baseline dense model. Code is available at Github.

cs.LG↗

arXiv2025

SpikePool: Event-driven Spiking Transformer with Pooling Attention

Building on the success of transformers, Spiking Neural Networks (SNNs) have increasingly been integrated with transformer architectures, leading to spiking transformers that demonstrate promising performance on event-based vision tasks. However, despite these empirical successes, there remains limited understanding of how spiking transformers fundamentally process event-based data. Current approaches primarily focus on architectural modifications without analyzing the underlying signal processing characteristics. In this work, we analyze spiking transformers through the frequency spectrum domain and discover that they behave as high-pass filters, contrasting with Vision Transformers (ViTs) that act as low-pass filters. This frequency domain analysis reveals why certain designs work well for event-based data, which contains valuable high-frequency information but is also sparse and noisy. Based on this observation, we propose SpikePool, which replaces spike-based self-attention with max pooling attention, a low-pass filtering operation, to create a selective band-pass filtering effect. This design preserves meaningful high-frequency content while capturing critical features and suppressing noise, achieving a better balance for event-based data processing. Our approach demonstrates competitive results on event-based datasets for both classification and object detection tasks while significantly reducing training and inference time by up to 42.5% and 32.8%, respectively.

cs.NE↗

arXiv2025

GS-Scale: Unlocking Large-Scale 3D Gaussian Splatting Training via Host Offloading

The advent of 3D Gaussian Splatting has revolutionized graphics rendering by delivering high visual quality and fast rendering speeds. However, training large-scale scenes at high quality remains challenging due to the substantial memory demands required to store parameters, gradients, and optimizer states, which can quickly overwhelm GPU memory. To address these limitations, we propose GS-Scale, a fast and memory-efficient training system for 3D Gaussian Splatting. GS-Scale stores all Gaussians in host memory, transferring only a subset to the GPU on demand for each forward and backward pass. While this dramatically reduces GPU memory usage, it requires frustum culling and optimizer updates to be executed on the CPU, introducing slowdowns due to CPU's limited compute and memory bandwidth. To mitigate this, GS-Scale employs three system-level optimizations: (1) selective offloading of geometric parameters for fast frustum culling, (2) parameter forwarding to pipeline CPU optimizer updates with GPU computation, and (3) deferred optimizer update to minimize unnecessary memory accesses for Gaussians with zero gradients. Our extensive evaluations on large-scale datasets demonstrate that GS-Scale significantly lowers GPU memory demands by 3.3-5.6x, while achieving training speeds comparable to GPU without host offloading. This enables large-scale 3D Gaussian Splatting training on consumer-grade GPUs; for instance, GS-Scale can scale the number of Gaussians from 4 million to 18 million on an RTX 4070 Mobile GPU, leading to 23-35% LPIPS (learned perceptual image patch similarity) improvement.

cs.CV↗

arXiv2025

AMRG: Extend Vision Language Models for Automatic Mammography Report Generation

Mammography report generation is a critical yet underexplored task in medical AI, characterized by challenges such as multiview image reasoning, high-resolution visual cues, and unstructured radiologic language. In this work, we introduce AMRG (Automatic Mammography Report Generation), the first end-to-end framework for generating narrative mammography reports using large vision-language models (VLMs). Building upon MedGemma-4B-it-a domain-specialized, instruction-tuned VLM-we employ a parameter-efficient fine-tuning (PEFT) strategy via Low-Rank Adaptation (LoRA), enabling lightweight adaptation with minimal computational overhead. We train and evaluate AMRG on DMID, a publicly available dataset of paired high-resolution mammograms and diagnostic reports. This work establishes the first reproducible benchmark for mammography report generation, addressing a longstanding gap in multimodal clinical AI. We systematically explore LoRA hyperparameter configurations and conduct comparative experiments across multiple VLM backbones, including both domain-specific and general-purpose models under a unified tuning protocol. Our framework demonstrates strong performance across both language generation and clinical metrics, achieving a ROUGE-L score of 0.5691, METEOR of 0.6152, CIDEr of 0.5818, and BI-RADS accuracy of 0.5582. Qualitative analysis further highlights improved diagnostic consistency and reduced hallucinations. AMRG offers a scalable and adaptable foundation for radiology report generation and paves the way for future research in multimodal medical AI.

eess.IV↗

arXiv2025

Quantitative pointwise estimates of the cooling process for inelastic Boltzmann equation

In this paper, we study the homogeneous inelastic Boltzmann equation for hard spheres. We first prove that the solution $f(t,v)$ is bounded pointwise from above by $C_{f_0}\langle t \rangle^3$ and establish that the cooling time is infinite $T_c = +\infty$ under the condition $f_0 \in L^1_2 \cap L^{\infty}_{s}$ for $s > 2$. Away from zero velocity, we further prove that $f(t,v)\leq C_{f_0, |v|} \langle t \rangle$ for $v \neq 0$ at any time $t > 0$. This time-dependent pointwise upper bound is natural in the cooling process, as we expect the density near $v = 0$ to grow rapidly. We also establish an upper bound that depends on the coefficient of normal restitution constant, $α\in (0,1]$. This upper bound becomes constant when $α= 1$, restoring the known upper bound for elastic collisions [8]. Consequently, through these results, we obtain Maxwellian upper bounds on the solutions at each time.

math.AP↗

arXiv2025

Global solutions in $L^{p}_{v}L^{\infty}_{x}$ for the Boltzmann equation in bounded domains

The existence theory for solutions to the Boltzmann equation in bounded domains has primarily been developed within uniformly bounded function classes, such as $L^{\infty}_{x,v}$, as in [Duan-Huang-Wang-Yang,2017], [Duan-Wang,2019], [Guo,2010]. In this paper, we investigate solutions in relaxed function spaces $L^{p}_{v}L^\infty_{x}$ for the initial-boundary value problem of the Boltzmann equation in bounded domains. We consider the case of hard potential under diffuse reflection boundary conditions and assume cutoff model. For large initial data in a weighted $L^{p}_{v}L^\infty_{x}$ space with small relative entropy, we construct unique global-in-time mild solution that converge exponentially to the global Maxwellian. A pointwise estimate for the gain term, bounded in terms of $L^p_v$ and $L^2_v$ norms, is essential to prove our main results. Relative to [Gualdani-Mischler-Mouhot,2017], our work provides an alternative perspective on convergence to equilibrium in the presence of boundary conditions.

math.AP↗

arXiv2025

FastPoint: Accelerating 3D Point Cloud Model Inference via Sample Point Distance Prediction

Deep neural networks have revolutionized 3D point cloud processing, yet efficiently handling large and irregular point clouds remains challenging. To tackle this problem, we introduce FastPoint, a novel software-based acceleration technique that leverages the predictable distance trend between sampled points during farthest point sampling. By predicting the distance curve, we can efficiently identify subsequent sample points without exhaustively computing all pairwise distances. Our proposal substantially accelerates farthest point sampling and neighbor search operations while preserving sampling quality and model performance. By integrating FastPoint into state-of-the-art 3D point cloud models, we achieve 2.55x end-to-end speedup on NVIDIA RTX 3090 GPU without sacrificing accuracy.

cs.CV↗

arXiv2025

FaithfulSAE: Towards Capturing Faithful Features with Sparse Autoencoders without External Dataset Dependencies

Sparse Autoencoders (SAEs) have emerged as a promising solution for decomposing large language model representations into interpretable features. However, Paulo and Belrose (2025) have highlighted instability across different initialization seeds, and Heap et al. (2025) have pointed out that SAEs may not capture model-internal features. These problems likely stem from training SAEs on external datasets - either collected from the Web or generated by another model - which may contain out-of-distribution (OOD) data beyond the model's generalisation capabilities. This can result in hallucinated SAE features, which we term "Fake Features", that misrepresent the model's internal activations. To address these issues, we propose FaithfulSAE, a method that trains SAEs on the model's own synthetic dataset. Using FaithfulSAEs, we demonstrate that training SAEs on less-OOD instruction datasets results in SAEs being more stable across seeds. Notably, FaithfulSAEs outperform SAEs trained on web-based datasets in the SAE probing task and exhibit a lower Fake Feature Ratio in 5 out of 7 models. Overall, our approach eliminates the dependency on external datasets, advancing interpretability by better capturing model-internal features while highlighting the often neglected importance of SAE training datasets.

cs.LG↗

arXiv2025

On the large amplitude solution of the Boltzmann equation with large external potential and boundary effects

The Boltzmann equation is a fundamental equation in kinetic theory that describes the motion of rarefied gases. In this study, we examine the Boltzmann equation within a $C^1$ bounded domain, subject to a large external potential $Φ(x)$ and diffuse reflection boundary conditions. Initially, we prove the asymptotic stability of small perturbations near the local Maxwellian $μ_E (x,v)$. Subsequently, we demonstrate the asymptotic stability of large amplitude solutions with initial data that is arbitrarily large in (weighted) $L^\infty$, but sufficiently small in the sense of relative entropy. Specifically, we extend the results for large amplitude solutions of the Boltzmann equation (with or without external potential) [10, 11, 12, 23] to scenarios involving significant external potentials [19, 28] under diffuse reflection boundary conditions.

math.AP↗

arXiv2025

GPTAQ: Efficient Finetuning-Free Quantization for Asymmetric Calibration

We introduce GPTAQ, a novel finetuning-free quantization method for compressing large-scale transformer architectures. Unlike the previous GPTQ method, which independently calibrates each layer, we always match the quantized layer's output to the exact output in the full-precision model, resulting in a scheme that we call asymmetric calibration. Such a scheme can effectively reduce the quantization error accumulated in previous layers. We analyze this problem using optimal brain compression to derive a close-formed solution. The new solution explicitly minimizes the quantization error as well as the accumulated asymmetry error. Furthermore, we utilize various techniques to parallelize the solution calculation, including channel parallelization, neuron decomposition, and Cholesky reformulation for matrix fusion. As a result, GPTAQ is easy to implement, simply using 20 more lines of code than GPTQ but improving its performance under low-bit quantization. Remarkably, on a single GPU, we quantize a 405B language transformer as well as EVA-02, the rank first vision transformer that achieves 90% pretraining Imagenet accuracy. Code is available at Github.

cs.LG↗

arXiv2025

Do Your Best and Get Enough Rest for Continual Learning

According to the forgetting curve theory, we can enhance memory retention by learning extensive data and taking adequate rest. This means that in order to effectively retain new knowledge, it is essential to learn it thoroughly and ensure sufficient rest so that our brain can memorize without forgetting. The main takeaway from this theory is that learning extensive data at once necessitates sufficient rest before learning the same data again. This aspect of human long-term memory retention can be effectively utilized to address the continual learning of neural networks. Retaining new knowledge for a long period of time without catastrophic forgetting is the critical problem of continual learning. Therefore, based on Ebbinghaus' theory, we introduce the view-batch model that adjusts the learning schedules to optimize the recall interval between retraining the same samples. The proposed view-batch model allows the network to get enough rest to learn extensive knowledge from the same samples with a recall interval of sufficient length. To this end, we specifically present two approaches: 1) a replay method that guarantees the optimal recall interval, and 2) a self-supervised learning that acquires extensive knowledge from a single training sample at a time. We empirically show that these approaches of our method are aligned with the forgetting curve theory, which can enhance long-term memory. In our experiments, we also demonstrate that our method significantly improves many state-of-the-art continual learning methods in various protocols and scenarios. We open-source this project at https://github.com/hankyul2/ViewBatchModel.

cs.CV↗

arXiv2025

Do Vision Models Develop Human-Like Progressive Difficulty Understanding?

When a human undertakes a test, their responses likely follow a pattern: if they answered an easy question $(2 \times 3)$ incorrectly, they would likely answer a more difficult one $(2 \times 3 \times 4)$ incorrectly; and if they answered a difficult question correctly, they would likely answer the easy one correctly. Anything else hints at memorization. Do current visual recognition models exhibit a similarly structured learning capacity? In this work, we consider the task of image classification and study if those models' responses follow that pattern. Since real images aren't labeled with difficulty, we first create a dataset of 100 categories, 10 attributes, and 3 difficulty levels using recent generative models: for each category (e.g., dog) and attribute (e.g., occlusion), we generate images of increasing difficulty (e.g., a dog without occlusion, a dog only partly visible). We find that most of the models do in fact behave similarly to the aforementioned pattern around 80-90% of the time. Using this property, we then explore a new way to evaluate those models. Instead of testing the model on every possible test image, we create an adaptive test akin to GRE, in which the model's performance on the current round of images determines the test images in the next round. This allows the model to skip over questions too easy/hard for itself, and helps us get its overall performance in fewer steps.

cs.CV↗