Institution profile

University of Windsor

Academic institutionnorthamerica · ca
Official website
Research library26linked papers
Opportunities0open roles
Selected work

Representative Papers

ResNetVLLM -- Multi-modal Vision LLM for the Video Understanding Task

Apr 20, 2025

For zero-shot video understanding, this paper proposes a cross-modal framework that eliminates the need for pre-trained video models. The method employs an off-the-shelf, non-video-pretrained ResNet as the visual encoder, directly interfaced with a large language model (LLM), and performs end-to-end joint optimization to align visual and linguistic representations. Its core contribution lies in departing from conventional video-specific pretraining paradigms: it pioneers the use of a frozen ResNet backbone—unfined on video data—combined with zero-shot prompt learning and cross-modal feature mapping, achieving superior generalization while preserving architectural simplicity. Experiments demonstrate state-of-the-art zero-shot performance on four standard benchmarks: MSRVTT-QA, MSVD-QA, TGIF-QA FrameQA, and ActivityNet-QA.

1 citations1 influentialRead paper

SMTpip: Interpreter-Aware SMT-Based Dependency Conflict Resolution for Restoring Python Source-Code Executability

Aug 16, 2026

This study addresses version conflicts, interpreter incompatibilities, and inefficient backtracking in Python dependency resolution by constructing a PyPI dependency knowledge graph and proposing an interpreter-aware SMT reasoning technique. By jointly encoding package dependencies and interpreter constraints into SMT formulas, this approach overcomes the limitations of traditional blind search methods, enabling precise co-resolution of dependencies and runtime environments. Experimental results demonstrate that the proposed method achieves speedups of 6.9× and 9.6× over pip and Conda, respectively. Furthermore, it consistently generates constraint-consistent executable environments, significantly enhancing both the efficiency and reliability of dependency resolution in complex Python ecosystems.

0 citationsRead paper

KyrgyzLLM-Bench: Benchmarking Kyrgyz Language Understanding

Jul 19, 2026

This study addresses the absence of native evaluation benchmarks for Kyrgyz-language large language models, as existing cross-lingual assessments predominantly rely on translated data and fail to capture the language’s unique linguistic and cultural characteristics. To bridge this gap, we introduce KyrgyzLLM-Bench, the first large-scale native benchmark for Kyrgyz, comprising the native datasets KyrgyzMMLU and KyrgyzRC, complemented by human-verified translation tasks. We systematically evaluate 26 open- and closed-source models under zero-shot and few-shot settings. Our analysis reveals that translation-induced plausibility shifts can compromise evaluation reliability, with consistent English–Kyrgyz performance on WinoGrande and BoolQ but a notable discrepancy on HellaSwag. Few-shot prompting enhances reading comprehension for certain open-source models. All data, code, and results are publicly released and integrated into mainstream multilingual evaluation frameworks.

0 citationsRead paper

Spectral-Morphological Attention U-Net: An Efficient Network for Active Wildfire Detection

Jul 17, 2026

This study addresses the challenge of early and accurate detection of active wildfires in satellite imagery by proposing a U-Net architecture that integrates spectral and morphological attention mechanisms. The method introduces, for the first time, a differentiable morphological gating mechanism, which is combined with a spectral attention module and a channel–spatial modulator and embedded within a residual attention U-Net backbone. This design significantly enhances the model’s sensitivity to wildfire-specific features and improves robustness in complex scenes. Evaluated on the TS-SatFire and Sen2Fire datasets, the proposed model achieves intersection-over-union (IoU) scores of 75.16% and 22.50%, respectively, outperforming existing approaches and demonstrating the effectiveness and novelty of the framework.

0 citationsRead paper
Recent publications

Latest Papers

SMTpip: Interpreter-Aware SMT-Based Dependency Conflict Resolution for Restoring Python Source-Code Executability

Aug 16, 2026

This study addresses version conflicts, interpreter incompatibilities, and inefficient backtracking in Python dependency resolution by constructing a PyPI dependency knowledge graph and proposing an interpreter-aware SMT reasoning technique. By jointly encoding package dependencies and interpreter constraints into SMT formulas, this approach overcomes the limitations of traditional blind search methods, enabling precise co-resolution of dependencies and runtime environments. Experimental results demonstrate that the proposed method achieves speedups of 6.9× and 9.6× over pip and Conda, respectively. Furthermore, it consistently generates constraint-consistent executable environments, significantly enhancing both the efficiency and reliability of dependency resolution in complex Python ecosystems.

0 citationsRead paper

KyrgyzLLM-Bench: Benchmarking Kyrgyz Language Understanding

Jul 19, 2026

This study addresses the absence of native evaluation benchmarks for Kyrgyz-language large language models, as existing cross-lingual assessments predominantly rely on translated data and fail to capture the language’s unique linguistic and cultural characteristics. To bridge this gap, we introduce KyrgyzLLM-Bench, the first large-scale native benchmark for Kyrgyz, comprising the native datasets KyrgyzMMLU and KyrgyzRC, complemented by human-verified translation tasks. We systematically evaluate 26 open- and closed-source models under zero-shot and few-shot settings. Our analysis reveals that translation-induced plausibility shifts can compromise evaluation reliability, with consistent English–Kyrgyz performance on WinoGrande and BoolQ but a notable discrepancy on HellaSwag. Few-shot prompting enhances reading comprehension for certain open-source models. All data, code, and results are publicly released and integrated into mainstream multilingual evaluation frameworks.

0 citationsRead paper

Spectral-Morphological Attention U-Net: An Efficient Network for Active Wildfire Detection

Jul 17, 2026

This study addresses the challenge of early and accurate detection of active wildfires in satellite imagery by proposing a U-Net architecture that integrates spectral and morphological attention mechanisms. The method introduces, for the first time, a differentiable morphological gating mechanism, which is combined with a spectral attention module and a channel–spatial modulator and embedded within a residual attention U-Net backbone. This design significantly enhances the model’s sensitivity to wildfire-specific features and improves robustness in complex scenes. Evaluated on the TS-SatFire and Sen2Fire datasets, the proposed model achieves intersection-over-union (IoU) scores of 75.16% and 22.50%, respectively, outperforming existing approaches and demonstrating the effectiveness and novelty of the framework.

0 citationsRead paper

An Objective Performance Evaluation of the LSTM Networks in Time Series Classification

May 18, 2026

This study addresses the lack of systematic evaluation comparing data-driven models, such as Long Short-Term Memory (LSTM) networks, with model-based approaches in structured time series classification tasks. The authors construct a controlled evaluation framework using Monte Carlo simulations to compare LSTM against an Expectation-Maximization (EM)-based classifier within linear Gaussian state-space models, across varying task difficulties, sequence lengths, and training set sizes. The theoretical performance upper bound is established by the likelihood ratio test derived from Kalman filter outputs under known model parameters. For the first time in a structured setting, this work quantifies the performance gap between these paradigms, revealing that LSTM exhibits performance saturation when classification relies solely on differences in measurement noise, failing to surpass the theoretical optimum even with increased data or longer sequences. In contrast, the EM-based classifier consistently approaches the upper bound when model assumptions hold.

0 citationsRead paper