Discrete Cosine Transform Based Causal Convolutional Neural Network for Drift Compensation in Chemical Sensors

TL;DR

This paper introduces a causal CNN with a DCT layer for effective drift compensation in chemical sensors, improving accuracy and robustness in noisy conditions.

Contribution

It proposes a novel DCT-based causal CNN architecture with learned soft-thresholding for denoising and estimating sensor drift signals.

Findings

Accurately estimates slowly varying drift signals

Effective in noisy sensor data

Outperforms traditional methods

Abstract

Sensor drift is a major problem in chemical sensors that requires addressing for reliable and accurate detection of chemical analytes. In this paper, we develop a causal convolutional neural network (CNN) with a Discrete Cosine Transform (DCT) layer to estimate the drift signal. In the DCT module, we apply soft-thresholding nonlinearity in the transform domain to denoise the data and obtain a sparse representation of the drift signal. The soft-threshold values are learned during training. Our results show that DCT layer-based CNNs are able to produce a slowly varying baseline drift signal. We train the CNN on synthetic data and test it on real chemical sensor data. Our results show that we can have an accurate and smooth drift estimate even when the observed sensor signal is very noisy.