arXiv subjects

Saiedeh Akbari

Publications and source records attributed to Saiedeh Akbari.

5 recordsLinked to original sources

arXiv2026

Effective Model Pruning: Measure The Redundancy of Model Components

This article initiates the study of a basic question about model pruning. Given a vector $s$ of importance scores assigned to model components, how many of the scored components could be discarded without sacrificing performance? We propose Effective Model Pruning (EMP), which derives the desired sparsity directly from the score distribution using the notion of effective sample size from particle filtering, also known as the inverse Simpson index. Rather than prescribe a pruning criterion, EMP supplies a universal adaptive threshold derived from the distribution of the score $s$ over the model components: EMP maps $s$ to a number $N_{eff}=N_{eff}(s)$, called the effective sample size. The $N-N_{eff}$ lowest scoring components are discarded. A tight lower bound on the effective mass $s_{eff}$ (the sum of retained normalized scores) in terms of $N_{eff}$ is derived. This process yields models with a provable upper bound on the loss change relative to the original dense model. Numerical experiments are performed demonstrating this phenomenon across a variety of network architectures including MLPs, CNNs, Transformers, LLMs, and KAN. It is also shown that EMP addresses a rich set of pruning criteria such as weight magnitude, attention score, KAN importance score, and even feature-level signals such as image pixels.

cs.LG↗

arXiv2025

Lyapunov-based Adaptive Transformer (LyAT) for Control of Stochastic Nonlinear Systems

This paper presents a novel Lyapunov-based Adaptive Transformer (LyAT) controller for stochastic nonlinear systems. While transformers have shown promise in various control applications due to sequential modeling through self-attention mechanisms, they have not been used within adaptive control architectures that provide stability guarantees. Existing transformer-based approaches for control rely on offline training with fixed weights, resulting in open-loop implementations that lack real-time adaptation capabilities and stability assurances. To address these limitations, a continuous LyAT controller is developed that adaptively estimates drift and diffusion uncertainties in stochastic dynamical systems without requiring offline pre-training. A key innovation is the analytically derived adaptation law constructed from a Lyapunov-based stability analysis, which enables real-time weight updates while guaranteeing probabilistic uniform ultimate boundedness of tracking and parameter estimation errors. Experimental validation on a quadrotor demonstrates the performance of the developed controller.

eess.SY↗

arXiv2025

LyLA-Therm: Lyapunov-based Langevin Adaptive Thermodynamic Neural Network Controller

Thermodynamic principles can be employed to design parameter update laws that address challenges such as the exploration vs. exploitation dilemma. In this paper, inspired by the Langevin equation, an update law is developed for a Lyapunov-based DNN control method, taking the form of a stochastic differential equation. The drift term is designed to minimize the system's generalized internal energy, while the diffusion term is governed by a user-selected generalized temperature law, allowing for more controlled fluctuations. The minimization of generalized internal energy in this design fulfills the exploitation objective, while the temperature-based stochastic noise ensures sufficient exploration. Using a Lyapunov-based stability analysis, the proposed Lyapunov-based Langevin Adaptive Thermodynamic (LyLA-Therm) neural network controller achieves probabilistic convergence of the tracking and parameter estimation errors to an ultimate bound. Simulation results demonstrate the effectiveness of the proposed approach, with the LyLA-Therm architecture achieving up to 20.66% improvement in tracking errors, up to 20.89% improvement in function approximation errors, and up to 11.31% improvement in off-trajectory function approximation errors compared to the baseline deterministic approach.

eess.SY↗

arXiv2024

Lyapunov-Based Deep Neural Networks for Adaptive Control of Stochastic Nonlinear Systems

Controlling nonlinear stochastic dynamical systems involves substantial challenges when the dynamics contain unknown and unstructured nonlinear state-dependent terms. For such complex systems, deep neural networks can serve as powerful black box approximators for the unknown drift and diffusion processes. Recent developments construct Lyapunov-based deep neural network (Lb-DNN) controllers to compensate for deterministic uncertainties using adaptive weight update laws derived from a Lyapunov-based analysis based on insights from the compositional structure of the DNN architecture. However, these Lb-DNN controllers do not account for non-deterministic uncertainties. This paper develops Lb-DNNs to adaptively compensate for both the drift and diffusion uncertainties of nonlinear stochastic dynamic systems. Through a Lyapunov-based stability analysis, a DNN-based approximation and corresponding DNN weight adaptation laws are constructed to eliminate the unknown state-dependent terms resulting from the nonlinear diffusion and drift processes. The tracking error is shown to be uniformly ultimately bounded in probability. Simulations are performed on a nonlinear stochastic dynamical system to show efficacy of the proposed method.

eess.SY↗

arXiv2023

Lyapunov-Based Dropout Deep Neural Network (Lb-DDNN) Controller

Deep neural network (DNN)-based adaptive controllers can be used to compensate for unstructured uncertainties in nonlinear dynamic systems. However, DNNs are also very susceptible to overfitting and co-adaptation. Dropout regularization is an approach where nodes are randomly dropped during training to alleviate issues such as overfitting and co-adaptation. In this paper, a dropout DNN-based adaptive controller is developed. The developed dropout technique allows the deactivation of weights that are stochastically selected for each individual layer within the DNN. Simultaneously, a Lyapunov-based real-time weight adaptation law is introduced to update the weights of all layers of the DNN for online unsupervised learning. A non-smooth Lyapunov-based stability analysis is performed to ensure asymptotic convergence of the tracking error. Simulation results of the developed dropout DNN-based adaptive controller indicate a 38.32% improvement in the tracking error, a 53.67% improvement in the function approximation error, and 50.44% lower control effort when compared to a baseline adaptive DNN-based controller without dropout regularization.

eess.SY↗