New Hybrid Neural Network Boosts AI Energy Efficiency

Scientists have created a hybrid convolutional neural network that incorporates oscillatory retinal neurons in its initial layer. This innovation results in a significantly more energy-efficient machine learning method. The self-oscillating circuit operates without external voltage and achieves performance comparable to advanced GPU-based CNNs, but with drastically reduced energy consumption.