‍ 🎯 Deep Learning and Neural Networks Symposium and Workshop

👨🏻‍🎓 Speaker introduction:

Dr. Timothée Masquelier,
CNRS Researcher (CR1) in Computational Neuroscience

Title: supervised learning in spiking neural networks

Abstract: I will present two recent works on supervised learning in spiking neural networks.
In the first one, we used backpropagation through time. The most commonly used spiking neuron model, the leaky integrate-and-fire neuron, obeys a differential equation which can be approximated using discrete time steps, leading to a recurrent relation for the potential. The firing threshold causes optimization issues, but they can be overcome using a surrogate gradient. We extended previous approaches in two ways. Firstly, we showed that the approach can be used to train convolutional layers. Secondly, we included fast horizontal connections à la Denève: when a neuron N fires, we subtract to the potentials of all the neurons with the same receptive the dot product between their weight vectors and the one of neuron N. Such connections improved the performance.
The second project focuses on SNNs which use at most one spike per neuron per stimulus, and latency coding. We derived a new learning rule for this sort of network, termed S4NN, akin to traditional error backpropagation, yet based on latencies. We show how approximate error gradients can be computed backward in a feedforward network with any number of layers.

⭕️ Check our website for more information

⚙️ Organizers: Institute for Cognitive and Brain Sciences, Shahid Beheshti University and Loop Academy


📢 @LoopAcademy
📢 @CMPLab

🌐 www.loopacademy.ir
🌐 www.cmplab.ir