Efficient Unstructured Pruning of Mamba State-Space Models for Resource-Constrained Environments

State-space models (SSMs), particularly the Mamba architecture, have emerged as powerful alternatives to Transformers for sequence modeling, offering linear-time complexity and competitive performance across diverse tasks. However, their large parameter counts pose significant challenges for deployment in resource-constrained environments. We propose a novel unstructured pruning framework tailored for Mamba models that achieves up to 70\% parameter reduction while retaining over 95\% of the original performance. Our approach integrates three key innovations: (1) a gradient-aware magnitude pruning technique that combines weight magnitude and gradient information to identify less critical parameters, (2) an iterative pruning schedule that gradually increases sparsity to maintain model stability, and (3) a global pruning strategy that optimizes parameter allocation across the entire model. Through extensive experiments on WikiText-103, Long Range Arena, and ETT time-series benchmarks, we demonstrate significant efficiency gains with minimal performance degradation. Our analysis of pruning effects on Mamba's components reveals critical insights into the architecture's redundancy and robustness, enabling practical deployment in resource-constrained settings while broadening Mamba's applicability.

Cache-Efficient Posterior Sampling for Reinforcement Learning with LLM-Derived Priors Across Discrete and Continuous Domains

Integrating large language models (LLMs) as priors in reinforcement learning (RL) offers significant advantages but comes with substantial computational costs. We present a principled cache-efficient framework for posterior sampling with LLM-derived priors that dramatically reduces these costs while maintaining high performance. At the core of our approach is an adaptive caching mechanism, where cache parameters are meta-optimized using surrogate gradients derived from policy performance. This design enables efficient inference across both discrete text environments (e.g., TextWorld, ALFWorld) and continuous control domains (e.g., MuJoCo), achieving a 3.8--4.7$\times$ reduction in LLM queries and 4.0--12.0$\times$ lower median latencies (85--93\,ms on a consumer GPU) while retaining 96--98\% of uncached performance. Our theoretical analysis provides KL divergence bounds on approximation quality, validated empirically. The framework extends to offline RL, where our CQL-Prior variant improves performance by 14--29\% and reduces training time by 38--40\%. Extensive evaluations across a diverse suite of eight tasks demonstrate the generalizability and practical viability of LLM-guided RL in resource-constrained settings.

HMAE: Self-Supervised Few-Shot Learning for Quantum Spin Systems

Quantum machine learning for spin and molecular systems faces critical challenges of scarce labeled data and computationally expensive simulations. To address these limitations, we introduce Hamiltonian-Masked Autoencoding (HMAE), a novel self-supervised framework that pre-trains transformers on unlabeled quantum Hamiltonians, enabling efficient few-shot transfer learning. Unlike random masking approaches, HMAE employs a physics-informed strategy based on quantum information theory to selectively mask Hamiltonian terms based on their physical significance. Experiments on 12,500 quantum Hamiltonians (60% real-world, 40% synthetic) demonstrate that HMAE achieves 85.3% $\pm$ 1.5% accuracy in phase classification and 0.15 $\pm$ 0.02 eV MAE in ground state energy prediction with merely 10 labeled examples - a statistically significant improvement (p < 0.01) over classical graph neural networks (78.1% $\pm$ 2.1%) and quantum neural networks (76.8% $\pm$ 2.3%). Our method's primary advantage is exceptional sample efficiency - reducing required labeled examples by 3-5x compared to baseline methods - though we emphasize that ground truth values for fine-tuning and evaluation still require exact diagonalization or tensor networks. We explicitly acknowledge that our current approach is limited to small quantum systems (specifically limited to 12 qubits during training, with limited extension to 16-20 qubits in testing) and that, while promising within this regime, this size restriction prevents immediate application to larger systems of practical interest in materials science and quantum chemistry.

ClearVision: Leveraging CycleGAN and SigLIP-2 for Robust All-Weather Classification in Traffic Camera Imagery

Accurate weather classification from low-quality traffic camera imagery remains a challenging task, particularly under adverse nighttime conditions. In this study, we propose a scalable framework that combines generative domain adaptation with efficient contrastive learning to enhance classification performance. Using CycleGAN-based domain translation, we improve the quality of nighttime images, enabling better feature extraction by downstream models. While the baseline EVA-02 model employing CLIP-based contrastive loss achieves an overall accuracy of 96.55\%, it exhibits a significant performance gap between daytime (97.21\%) and nighttime conditions (63.40\%). Replacing CLIP with the lightweight SigLIP-2 (Sigmoid contrastive loss) achieves a competitive overall accuracy of 94.00\%, with substantial improvements in nighttime performance (85.90\% accuracy). The combination of Vision-SigLIP-2, Text-SigLIP-2, CycleGAN, and contrastive training achieves the best nighttime accuracy (85.90\%) among all models tested, while EVA-02 with CycleGAN maintains the highest overall accuracy (97.01\%) and per-class accuracies. These findings demonstrate the potential of combining domain adaptation and efficient contrastive learning to build practical, resource-efficient weather classification systems for intelligent transportation infrastructure.

Accuracy is Not Agreement: Expert-Aligned Evaluation of Crash Narrative Classification Models

This study explores the relationship between deep learning (DL) model accuracy and expert agreement in the classification of crash narratives. We evaluate five DL models -- including BERT variants, the Universal Sentence Encoder (USE), and a zero-shot classifier -- against expert-labeled data and narrative text. The analysis is further extended to four large language models (LLMs): GPT-4, LLaMA 3, Qwen, and Claude. Our results reveal a counterintuitive trend: models with higher technical accuracy often exhibit lower agreement with domain experts, whereas LLMs demonstrate greater expert alignment despite relatively lower accuracy scores. To quantify and interpret model-expert agreement, we employ Cohen's Kappa, Principal Component Analysis (PCA), and SHAP-based explainability techniques. Findings indicate that expert-aligned models tend to rely more on contextual and temporal language cues, rather than location-specific keywords. These results underscore that accuracy alone is insufficient for evaluating models in safety-critical NLP applications. We advocate for incorporating expert agreement as a complementary metric in model evaluation frameworks and highlight the promise of LLMs as interpretable, scalable tools for crash analysis pipelines.

Crash Time Matters: HybridMamba for Fine-Grained Temporal Localization in Traffic Surveillance Footage

Traffic crash detection in long-form surveillance videos is critical for emergency response and infrastructure planning but remains difficult due to the brief and rare nature of crash events. We introduce HybridMamba, a novel architecture that combines visual transformers with state-space temporal modeling to achieve accurate crash time localization. Our method uses multi-level token compression and hierarchical temporal processing to remain computationally efficient without sacrificing temporal resolution. Evaluated on a large-scale dataset from the Iowa Department of Transportation, HybridMamba achieves a mean absolute error of 1.50 seconds, with 65.2 percent of predictions within one second of the ground truth. It outperforms recent video-language models such as TimeChat and VideoLLaMA2 by up to 2.8 seconds, while using significantly fewer parameters. Our results demonstrate strong generalization across videos ranging from 2 to 40 minutes in diverse conditions. HybridMamba offers a robust and efficient solution for fine-grained temporal localization in traffic surveillance. The code will be released upon publication.

DeepLocalization: Using change point detection for Temporal Action Localization

In this study, we introduce DeepLocalization, an innovative framework devised for the real-time localization of actions tailored explicitly for monitoring driver behavior. Utilizing the power of advanced deep learning methodologies, our objective is to tackle the critical issue of distracted driving-a significant factor contributing to road accidents. Our strategy employs a dual approach: leveraging Graph-Based Change-Point Detection for pinpointing actions in time alongside a Video Large Language Model (Video-LLM) for precisely categorizing activities. Through careful prompt engineering, we customize the Video-LLM to adeptly handle driving activities' nuances, ensuring its classification efficacy even with sparse data. Engineered to be lightweight, our framework is optimized for consumer-grade GPUs, making it vastly applicable in practical scenarios. We subjected our method to rigorous testing on the SynDD2 dataset, a complex benchmark for distracted driving behaviors, where it demonstrated commendable performance-achieving 57.5% accuracy in event classification and 51% in event detection. These outcomes underscore the substantial promise of DeepLocalization in accurately identifying diverse driver behaviors and their temporal occurrences, all within the bounds of limited computational resources.