SPAYK
About Spayk
Robotics / Computer Vision
Spayk has been specifically proposed for researchers outside the field of neuroscience to accelerate the study of spiking neural networks.
Easy to Use
Spayk is an open source Python library that can be easily used. It has a clear syntax. Spiked spiking neural networks can be easily expressed and simulated.
Scalable
Spike can scale very well the increase in the number of neuron groups and synaptic connections.
GUI
Visualization developments continues.
A quick start to Spayk
We recommend creating a new virtual environment to use Spayk.
Experiments
neuron-groups
Classifier with An Izhikevich Neuron Group
In the first part of the experiment, a group of 250 neurons was defined. About 20 percent of the neurons in the input stimuli were selected to form inhibitory synapses with neurons in the Izhikevich group. Each neuron in the Izhikevich group was randomly connected to 70 percent of the neurons in the stimuli.
read more
Unsupervised-learning
Unsupervised Pattern Recognition
The experiment from Masquelier et al.’s study was performed to bring the spike response model simulation capability to the SPAYK environment. It was demonstrated in the study that a neuron with a spike response model can detect repeating patterns in continuous spike trains by modifying weights via STDP. To replicate the study’s findings, a single SRM LIF neuron with STDP capability was stimulated for 15 s with a special Poisson spike train of 2000 neurons
read more
supervised-learning
Supervised MNIST Classifier
In this experiment, 3 random samples labeled 1, 4, and 7 from the MNIST dataset were converted into 500 ms long Poisson spike trains. To generate stimuli, these spike trains were arranged to repeat twice at random times.
read moreDevelopers
Aykut Görkem GELEN
Developer
SPAYK is a simulation environment for spiking neural networks.
Ayten ATASOY
Advisor
Karadeniz Technical University, Department of Electrical and Electronics Engineering