leamoon/StochasticNet
Nerual Network of Stochastic Computing for MNIST Recognition
Implements LeNet-5 architecture using stochastic computing instead of full-precision arithmetic, reducing circuit complexity for a Minecraft redstone CNN that recognizes 15×15 handwritten digits with 80% accuracy. Stochastic encoding trades computation speed (~20 minutes per inference) for dramatically simpler physical layout compared to traditional digital logic. Demonstrates feasibility of probabilistic neural networks in resource-constrained environments, with potential applications to analog and neuromorphic hardware designs.
225 stars. No commits in the last 6 months.
Stars
225
Forks
11
Language
Python
License
MIT
Category
Last pushed
Jul 01, 2022
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/leamoon/StochasticNet"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
Jasonnor/Backpropagation
Implementing multilayer neural networks through backpropagation using Java.
dlidstrom/NeuralNetworkInAllLangs
Vanilla neural network implemented in all major languages
goessl/MachineLearning
An easy neural network for Java!
thag0/JNN
Biblioteca para manipulação de modelos de Redes Neurais Artificiais com Java
evolvingstuff/RecurrentJava
Deep Recurrent Neural Nets in Java