Counter-Example Guided Neural Network Controller Synthesis

Title of the Talk: CCounter-Example Guided Neural Network Controller Synthesis
Speakers: Dr. Inzemamul Haque
Host Faculty: Dr.Ashish Mishra
Date: Jun 25, 2025
Time: 11:00am to 12:00 pm
Venue: CS-LH1

Abstract: We propose a framework for training neural networks (NNs) to imitate complex controllers (like Model Predictive Control (MPC)) for control requirements expressed using Signal Temporal Logic (STL). Initially, the NN is trained using data from grid based sampling of an expert controller (e.g., Proportional Integral Derivative (PID), MPC). The trained NN is evaluated against the behaviour specified in STL and retrained with data from the expert controller at states where the NN controller fails. This iterative training continues until there are no failures. Moreover, we introduce a method to evaluate the performance of the learned controller via parameterization and parameter estimation of the STL requirements. We demonstrate our approach with a non-linear flying robot system with MPC.

Bio: Inzemamul Haque is a Senior Lecturer at Krea University. He works on application of formal methods and artificial intelligence to software systems and cyber-physical systems. He completed his PhD from Indian Institute of Science, Bangalore, where he worked on the formal verification of an OS microkernel. After his PhD, Inzemamul Haque was a postdoctoral fellow at the Department of Computer Science and Engineering at IIT Kanpur, where he worked on counter-example guided synthesis of neural-network based controllers and action plan synthesis of collaborative robots. Inzemamul Haque holds a B.Tech. in Computer Engineering from Aligarh Muslim University.