What question did this study set out to answer?

The aim is to explore surrogate modeling in electronic circuits for optimizing functional problems using gradient-based methods.

March 23, 2026Open Access

Design of Analog Computers: Building Blocks and Advanced Methods for Tiny Analog Machine Learning with Surrogate Modeling

Key Points

The aim is to explore surrogate modeling in electronic circuits for optimizing functional problems using gradient-based methods.
Investigation of surrogate modeling for electronic circuits
Use of gradient-based optimization techniques
Demonstration with an Analog Artificial Neural Network
Evaluation using the IRIS benchmark dataset
Surrogate modeling is significantly more computationally expensive than pure mathematical models, roughly 10 times more.
Error gradient back-propagation showed instability and random convergence during training.
Classification error improved to 2% using the adam optimizer during testing with the IRIS dataset.
Training results are transferable to an analog transistor circuit without loss of accuracy.

Abstract

This study investigates the use of surrogate modeling of electronic circuits for functional optimization problems using gradient-based methods, demonstrated with an Analog Artificial Neural Network use-case. The results show that a surrogate model approach is more computationally expensive than a pure mathematical model (about 10 times), that the training process using error gradient back-propagation is instable and only randomly convergent, but it significantly improved classification error shown with the IRIS benchmark dataset, with the adam optimizer finding the "golden standard" solution with 2% classification error. The training results can be directly transferred to an analog transistor circuit without loss of accuracy.

Design of Analog Computers: Building Blocks and Advanced Methods for Tiny Analog Machine Learning with Surrogate Modeling

Key Points

Abstract

Cite This Study