Signed Spiking Neuron Enabled by an Orthogonal-Easy-Axis Magnetic Tunnel Junction
Researchers have developed a new type of neuron using a magnetic tunnel junction (MTJ) that can process signed information, offering richer data representation than standard spiking neurons. This MTJ-based neuron mimics the behavior of a signed leaky integrate-and-fire (LIF) model, enabling bipolar spike generation. When integrated into neural networks, this design achieved high accuracy on image recognition tasks like CIFAR-10, closely matching the performance of ideal signed LIF neurons. AI
IMPACT This novel neuron design could lead to more efficient and powerful neuromorphic computing systems for AI.