Abstract:
The cognitive workload plays a critical role in tasks that require dynamic decision-making and are
conducted under high-risk, real-time conditions. A high workload might result in unforeseen and
disproportionate dangers, whereas the result of a low workload is disengagement from the activity.
This emphasizes the significance of maintaining adequate cognitive effort in high-risk settings to
complete the task successfully. In this study, we employ several functional connectivity measurements
in conjunction with deep learning to categorize cognitive workload. This study makes use of an Nback
EEG dataset. Following pre-processing, functional connectivity features such as phase locking
value (PLV), phase lagging index (PLI), and coherency were extracted. These characteristics are
directed/non-directed, allowing for speedier calculations. The deep learning classifier CNN utilizes
these features to classify the cognitive workload into three categories: low (0-back), medium (2-back),
and high (3-back).we achieve the highest accuracy of 93.75 % using PLV in CNN-A architecture,
87.5% accuracy by using Coherency in CNN-A Architecture, and 68.75% using PLI in CNN-A
architecture. By leveraging deep learning methods to analyze functional connectivity features, this
research opens avenues for understanding and supporting cognitive processes in various domains,
including robotics, healthcare, and education. The scope of the study could be extended to explore the
possibility of categorizing cognitive effort in complex, real-world situations that occur in real time
and involve dynamic and intricate circumstances.
