A PhD dissertation in Electrical Engineering proposes energy-efficient architectural designs

A PhD dissertation from the College of Electrical Engineering at the University of Technology, panels (low-power techniques based on approximate and radial machine learning techniques) presented by student Iyad Mohammed Khorshed, reached that energy-efficient architectural designs for machine learning inference applications. Approximation techniques are applied at four abstraction levels: the data level, the circuit level, the architectural level, and the system level. Machine learning models are trained on high-dimensional datasets containing many useless, redundant, or noisy features, which leads to computational overhead and a deterioration in model accuracy.

The discussion committee consisted of:

• Dr. Raed Faleh Hassan / the head.
• Dr. Hadeel Nasrat Abdullah / Member.
• Dr. Mohammed Najm Abdullah / Member.
• Prof. Dr. Iyad Ibrahim Ali / Member.
• Prof. Dr. Ikhlas Kahdim Hamza / Member.
• Dr. Amjad Jalil Humidi / Supervisor.