KB
Kayhan Behdin
cs.LGstat.MLcs.AIcs.IRstat.MEcs.CLcs.CRcs.CVcs.MMcs.SD
On Valency
published · living versionsW_y66aybk3·v1 · currentpublished
Reconstruction-Optimized Expert Pruning for Mixture-of-Experts Language Models
with Jelena Markovic-Voronov, Petros Prastakos, Jincheng Cao, Zhipeng Wang +5
1 version
Preprints & journals
30 papers in the corpus · 2017–2026Reconstruction-Optimized Expert Pruning for Mixture-of-Experts Language Models2610.27123v1 · Jelena Markovic-Voronov, Petros Prastakos, Kayhan Behdin et al.2026 · 0 citationsarXivon Valency
Target-ESS Tempered Policy Optimization (TETPO): Smooth Variance Control via Bridge Importance Sampling2610.14440v1 · Shayan Mohajer Hamidi, Kayhan Behdin, Wenhui Zhu et al.2026 · 0 citationsarXiv
Smaller Models, Better Rejects: Preference Distillation Scaling2609.38987v1 · Rui Cai, Wenhui Zhu, Xiwen Chen et al.2026 · 0 citationsarXiv
TREK: Distill to Explore, Reinforce to Refine2607.05339v1 · Yuanda Xu, Zhengze Zhou, Kayhan Behdin et al.2026 · 0 citationsarXiv
Sampling for Quality: Training-Free Reward-Guided LLM Decoding via Sequential Monte Carlo2604.16453v1 · Jelena Markovic-Voronov, Wenhui Zhu, Bo Long et al.2026 · 0 citationsarXiv
Modeling with Categorical Features via Exact Fusion and Sparsity Regularisation2603.25854v1 · Kayhan Behdin, Riade Benbaki, Peter Radchenko et al.2026 · 0 citationsarXiv
Robust Batch-Level Query Routing for Large Language Models under Cost and Capacity Constraints2603.26796v1 · Jelena Markovic-Voronov, Kayhan Behdin, Yuanda Xu et al.2026 · 0 citationsarXiv
Semantic Search At LinkedIn2602.07309v1 · Fedor Borisyuk, Sriram Vasudevan, Muchen Wu et al.2026 · 0 citationsarXiv
Multi-task learning for sparsity pattern heterogeneity: statistical and computational perspectives.42459560 · Behdin, Kayhan, Loewinger, Gabriel, Kishida, Kenneth T et al.2026 · 0 citationsJournal of the Royal Statistical Society. Series B, Statistical methodology. 2026;88(3):1029-1059
MixLM: High-Throughput and Effective LLM Ranking via Text-Embedding Mix-Interaction2512.07846v2 · Guoyao Li, Ran He, Shusen Jing et al.2025 · 0 citationsarXiv
Scaling Up Efficient Small Language Models Serving and Deployment for Semantic Job Search2510.22101v1 · Kayhan Behdin, Qingquan Song, Sriram Vasudevan et al.2025 · 0 citationsarXiv
Differentially Private High-dimensional Variable Selection via Integer Programming2510.22062v1 · Petros Prastakos, Kayhan Behdin, Rahul Mazumder2025 · 0 citationsarXiv
Reasoning Models Can be Accurately Pruned Via Chain-of-Thought Reconstruction2509.12464v2 · Ryan Lucas, Kayhan Behdin, Zhipeng Wang et al.2025 · 0 citationsarXiv
BP-Seg: A graphical model approach to unsupervised and non-contiguous text segmentation using belief propagation2505.16965v2 · Fengyi Li, Kayhan Behdin, Natesh Pillai et al.2025 · 0 citationsarXiv
An Optimization Framework for Differentially Private Sparse Fine-Tuning2503.12822v1 · Mehdi Makni, Kayhan Behdin, Gabriel Afriat et al.2025 · 0 citationsarXiv
Scaling Down, Serving Fast: Compressing and Deploying Efficient LLMs for Recommendation Systems2502.14305v2 · Kayhan Behdin, Ata Fatahibaarzi, Qingquan Song et al.2025 · 1 citationarXiv
HASSLE-free: A unified Framework for Sparse plus Low-Rank Matrix Decomposition for LLMs2502.00899v1 · Mehdi Makni, Kayhan Behdin, Zheng Xu et al.2025 · 0 citationsarXiv
ALPS: Improved Optimization for Highly Sparse One-Shot Pruning for Large Language Models2406.07831v3 · Xiang Meng, Kayhan Behdin, Haoyue Wang et al.2024 · 1 citationarXiv
OSSCAR: One-Shot Structured Pruning in Vision and Language Models with Combinatorial Optimization2403.12983v1 · Xiang Meng, Shibal Ibrahim, Kayhan Behdin et al.2024 · 2 citationsarXiv
End-to-end Feature Selection Approach for Learning Skinny Trees2310.18542v3 · Shibal Ibrahim, Kayhan Behdin, Rahul Mazumder2023 · 0 citationsInternational Conference on Artificial Intelligence and Statistics (AISTATS) 2024
QuantEase: Optimization-based Quantization for Language Models2309.01885v2 · Kayhan Behdin, Ayan Acharya, Aman Gupta et al.2023 · 3 citationsarXiv
Sparse Gaussian Graphical Models with Discrete Optimization: Computational and Statistical Perspectives2307.09366v2 · Kayhan Behdin, Wenyu Chen, Rahul Mazumder2023 · 0 citationsarXiv
On Statistical Properties of Sharpness-Aware Minimization: Provable Guarantees2302.11836v3 · Kayhan Behdin, Rahul Mazumder2023 · 1 citationarXiv
mSAM: Micro-Batch-Averaged Sharpness-Aware Minimization2302.09693v2 · Kayhan Behdin, Qingquan Song, Aman Gupta et al.2023 · 1 citationarXiv
Improved Deep Neural Network Generalization Using m-Sharpness-Aware Minimization2212.04343v1 · Kayhan Behdin, Qingquan Song, Aman Gupta et al.2022 · 2 citationsarXiv
Sparse PCA: A New Scalable Estimator Based On Integer Programming2109.11142v4 · Kayhan Behdin, Rahul Mazumder2021 · 2 citationsarXiv
Sparse NMF with Archetypal Regularization: Computational and Robustness Properties2104.03527v2 · Kayhan Behdin, Rahul Mazumder2021 · 0 citationsarXiv
Recovering Quantized Data with Missing Information Using Bilinear Factorization and Augmented Lagrangian Method1810.03222v1 · Ashkan Esmaeili, Kayhan Behdin, Sina Al-E-Mohammad et al.2018 · 3 citationsarXiv
Transduction with Matrix Completion Using Smoothed Rank Function1805.07561v1 · Ashkan Esmaeili, Kayhan Behdin, Mohammad Amin Fakharian et al.2018 · 5 citationsarXiv
OBTAIN: Real-Time Beat Tracking in Audio Signals1704.02216v2 · Ali Mottaghi, Kayhan Behdin, Ashkan Esmaeili et al.2017 · 8 citationsarXiv
Career total: 41 works. 30 are in this corpus.
Profile built from the corpus for this byline.
Author records are still filling in while the Hub is in alpha. If this is your page, you'll be able to claim it soon. Spot a mistake? Tell us.