Chebyshev Hiking Optimization with Squeeze Excitation Depthwise Dilated ResNet based IT Professional Performance Analysis
P Yuvapriya, Subramanian P, R Surendran · 2025
While considering the digital industries, accurately evaluating the performance of IT professionals is critical for informed talent management and organizational growth. Traditional methods of performance analysis lack adaptability and fail to integrate heterogeneous data sources like structured metrics and unstructured self-assessments or feedback. To address this challenge, this research proposes an advanced classification framework that integrates natural language processing and deep learning optimization. Initially, performance-related data is acquired and is pre-processed for removing artefacts. Then, the outcome is passed through BERT-based feature extractor to generate semantically rich representations. The extracted feature vector is then fed into a novel Chebyshev Hiking Optimization-based Squeeze Excitation-enhanced Depthwise Dilated ResNet (CH-SE2DRNet) model. The SE2DRNet module captures both local and long-range dependencies using depthwise and dilated convolutions and the Squeeze-and-Excitation mechanism recalibrates feature importance dynamically. To enhance the accuracy of performance analysis, the Chebyshev Hiking (CH) Optimization algorithm is designed to fine-tune model parameters. The proposed system demonstrated superior with improvements in accuracy over traditional methods.