Sitemap
A list of all the posts and pages found on the site. For you robots out there, there is an XML version available for digesting as well.
Pages
Posts
notes
portfolio
Algorithm of Thoughts: Enhancing Exploration of Ideas in Large Language Models
Some tasks — like planning problems — cannot be solved linearly. Chain-of-Thought works for step-by-step reasoning but fails when you need to explore multiple paths. Tree-of-Thoughts solves this but requires many separate LLM queries per problem plus external code to manage the tree. AoT’s insight is simpler: show the model examples of complete search trajectories — including backtracking and dead ends — and it learns to internalize the search itself. No external tree management. One query.
A Dynamic Penalization Framework for Online Rank-1 Semidefinite Programming Relaxations
In Dynamic Penalization for Rank-1 SDP Relaxations (L4DC 2025, with Lavaei and Jin), we differentiate through a penalized SDP solver to learn penalty matrices that drive relaxations toward rank-1 solutions, and meta-learn initializations across tasks for faster, feasibility-preserving solves on Max-Cut and optimal power flow.
publications
A Singular Value Thresholding Algorithm for Order Estimation
Published in American Control Conference (ACC), 2023
A singular value thresholding approach for determining the order of linear dynamic systems.
Recommended citation: Ahmad Al-Tawaha, Khaled F. Aljanaideh, and A. Alshorman. "A Singular Value Thresholding Algorithm for Order Estimation." In *Proceedings of the American Control Conference (ACC)*, 2023.
Learning-to-Learn to Guide Random Search: Derivative-Free Meta Blackbox Optimization on Manifold
Published in 5th Annual Learning for Dynamics & Control Conference (L4DC), 2023
Meta-learning framework that learns to guide random search for derivative-free blackbox optimization over manifolds.
Recommended citation: Bilgehan Sel, Ahmad Al-Tawaha, Yuhao Ding, Ruoxi Jia, Bo Ji, Javad Lavaei, and Ming Jin. "Learning-to-Learn to Guide Random Search: Derivative-Free Meta Blackbox Optimization on Manifold." In *Proceedings of the Learning for Dynamics & Control Conference (L4DC)*, 2023.
Download Paper
Decision-Focused Learning for Inverse Noncooperative Games: Generalization Bounds and Convergence Analysis
Published in IFAC World Congress (IFAC-PapersOnLine), 2023
Inverse game learning method that embeds equilibrium computation into a differentiable decision-focused learning pipeline with generalization guarantees.
Recommended citation: Ahmad Al-Tawaha, Harshal Kaushik, Bilgehan Sel, Ruoxi Jia, and Ming Jin. "Decision-Focused Learning for Inverse Noncooperative Games: Generalization Bounds and Convergence Analysis." *IFAC-PapersOnLine*, 56(2):9336–9341, 2023.
Download Paper
Does Online Gradient Descent (and Variants) Still Work with Biased Gradient and Variance?
Published in American Control Conference (ACC), 2024
Analyzes online gradient descent and its variants under biased gradient estimates and non-stationary variance, characterizing when convergence guarantees still hold.
Recommended citation: Ahmad Al-Tawaha and Ming Jin. "Does Online Gradient Descent (and Variants) Still Work with Biased Gradient and Variance?" In *Proceedings of the American Control Conference (ACC)*, 2024.
Algorithm of Thoughts: Enhancing Exploration of Ideas in Large Language Models
Published in International Conference on Machine Learning (ICML), 2024
In-context algorithmic prompting strategy that guides LLMs along algorithm-like reasoning paths to improve math and general reasoning performance.
Recommended citation: Bilgehan Sel, Ahmad Al-Tawaha, Vanshaj Khattar, Ruoxi Jia, and Ming Jin. "Algorithm of Thoughts: Enhancing Exploration of Ideas in Large Language Models." In *Proceedings of the International Conference on Machine Learning (ICML)*, 2024.
Download Paper
A Nonheuristic Singular Value Thresholding Algorithm for Order Estimation
Published in Journal of Dynamic Systems, Measurement, and Control (ASME), 2025, 2025
A nonheuristic, analytically motivated singular-value thresholding algorithm for accurate model order determination in noisy dynamic systems.
Recommended citation: Ahmad Al-Tawaha, A. Alshorman, and Khaled F. Aljanaideh. "A Nonheuristic Singular Value Thresholding Algorithm for Order Estimation." Journal of Dynamic Systems, Measurement, and Control (ASME), vol. 147, no. 5, 2025.
Download Paper
Distributed Optimization and Learning: A Paradigm Shift for Power Systems
Published in IEEE Systems Journal (accepted, 2025), 2025
Survey and perspective on distributed optimization and learning for power systems.
Recommended citation: A. Al-Tawaha, E. Cibaku, S. Park, J. Lavaei, and M. Jin. "Distributed Optimization and Learning: A Paradigm Shift for Power Systems." IEEE Systems Journal, accepted, 2025.
Defense against Joint Poison and Evasion Attacks: A Case Study of DERMS
Published in AAAI 2025 Workshop, 2025
Case study on defending Distributed Energy Resource Management Systems (DERMS) against combined poisoning and evasion adversarial attacks.
Recommended citation: Ahmad Al-Tawaha*, Zain Ul-Abdeen*, Padmaksha Roy*, Ruoxi Jia, Laura Freeman, Peter Beling, Chen-Ching Liu, Alberto Sangiovanni-Vincentelli, and Ming Jin. "Defense against Joint Poison and Evasion Attacks: A Case Study of DERMS." *AAAI 2025 Workshop*.
Monte Carlo Grid Dynamic Programming: Almost Sure Convergence and Probability Constraints
Published in American Control Conference (ACC), 2025
Meshless Monte Carlo-based dynamic programming method with convergence guarantees and support for probabilistic state constraints.
Recommended citation: Mohammad S. Ramadan, Ahmad Al-Tawaha, Mohamed Shouman, Ahmed Atallah, and Ming Jin. "Monte Carlo Grid Dynamic Programming: Almost Sure Convergence and Probability Constraints." In *Proceedings of the American Control Conference (ACC)*, 2025.
Download Paper
An Analytical Approach to Signal Denoising Based on Singular Value Decomposition
Published in American Control Conference (ACC), 2025
Non-iterative signal denoising method using singular values of Hankel matrices built from noisy measurements.
Recommended citation: Ahmad Al-Tawaha, Khaled Aljanaideh, and coauthors. "An Analytical Approach to Signal Denoising Based on Singular Value Decomposition." In *Proceedings of the American Control Conference (ACC)*, 2025.
Download Paper
A Dynamic Penalization Framework for Online Rank-1 Semidefinite Programming Relaxations
Published in 7th Annual Learning for Dynamics & Control Conference (L4DC), 2025
Unified framework for solving sequences of SDP relaxations with rank-one constraints in online settings using dynamic penalization.
Recommended citation: Ahmad Al-Tawaha, Javad Lavaei, and Ming Jin. "A Dynamic Penalization Framework for Online Rank-1 Semidefinite Programming Relaxations." In *Proceedings of the 7th Annual Learning for Dynamics & Control Conference (L4DC)*, pp. 1012–1024, 2025.
Download Paper
Finite-Time Identification of LTI Systems Using Non-Causal FIR Models: A Unified Framework for Stable and Unstable Systems
Published in Under review, 2025
A unified framework for finite-time identification of both stable and unstable LTI systems via non-causal FIR models.
Recommended citation: Ahmad Al-Tawaha, Ming Jin, and Khaled F. Aljanaideh. "Finite-Time Identification of LTI Systems Using Non-Causal FIR Models: A Unified Framework for Stable and Unstable Systems." *Under review*.
Remembering More, Risking More: Longitudinal Safety Risks in Memory-Equipped LLM Agents
Published in Under review at NeurIPS, 2025
Studies how persistent memory amplifies safety risks over time in memory-equipped LLM agents, characterizing temporal amplification of memory-induced safety violations. Under review at NeurIPS.
Recommended citation: Ahmad Al-Tawaha, Ruoxi Jia, and Ming Jin. "Remembering More, Risking More: Longitudinal Safety Risks in Memory-Equipped LLM Agents." *Under review at NeurIPS*.