Mental workload classification from non-invasive fNIRs signals through deep convolutional neural network

Classification of mental workload has always been considered a crucial task in the literature related to brain mem-ory. People perform various tasks and have multiple cognitive workloads. This mental workload can be sensed in a non-intrusive way using Functional near-infrared spectroscopy (fNIRS) sig-nals. fNIRS is a photosensitive brain examining method which uses near-infrared spectroscopy to measure aspects of brain functions and activities. In this work, we focus on classifying segmented mental workload from fNIRS signals. We propose a deep convolutional neural (DCNN) network to classify mental workload. We evaluate our model performance using the publicly available large-scale open-access dataset, “Tufts fNIRS to Mental Workload (fNIRS2MW)” that consists of 68 participants per-forming n-back tasks where increased n represents the intensity of the mental workload. Our proposed deep convolutional neural network (DCNN) comprises six convolutional layers. Our DCNN achieves a performance gain of 28 % and 4 % comparing the state-of-the-art models EEGnet and Deep ConvNet, respectively.