neuromorphic-computing

micronet, a model compression and deploy lib. compression: 1、quantization: quantization-aware-training(QAT), High-Bit(>2b)(DoReFa/Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference)、Low-Bit(≤2b)/Ternary and Binary(TWN/BNN/XNOR-Net); post-training-quantization(PTQ), 8-bit(tensorrt); 2、 pruning: normal、regular and group convolutional channel pruning; 3、 group convolution structure; 4、batch-normalization fuse for quantization. deploy: tensorrt, fp32/fp16/int8(ptq-calibration)、op-adapt(upsample)、dynamic_shape

A Software Framework for Neuromorphic Computing

Actively developed Hierarchical Temporal Memory (HTM) community fork (continuation) of NuPIC. Implementation for C++ and Python

Long short-term memory Spiking Neural Networks

This repository will host models, modules, algorithms and applications developed by the INRC Community to run on the Intel Loihi Platform.

Deep learning for spiking neural networks

Learn about the Neumorphic engineering process of creating large-scale integration (VLSI) systems containing electronic analog circuits to mimic neuro-biological architectures.

Deep Learning library for Lava

Spiking-DDPG trains an SNN for energy-efficient mapless navigation on Intel's Loihi neuromorphic processor.

Constraint Optimization with Lava

Low-level Python APIs for Accessing Neuromorphic Devices.

Dynex is a next-generation platform for neuromorphic computing based on a new flexible blockchain protocol. It consists of participating nodes that together constitute one enormous neuromorphic computing network. Consequently, the platform is capable of performing computations at unprecedented speeds and efficiency – even exceeding quantum computing. Everyone is welcome to participate, since the Dynex neuromorphic computing chip is capable of being simulated using almost any device, from regular laptops to desktop computers to GPUs, FPGAs and ASIC clusters. Users exchange computation time for Dynex’s native token DNX, thus enabling everyone to earn money on the platform.

NeuroMorphic Predictive Model with Spiking Neural Networks (SNN) using Pytorch

Structured clustering for memristive crossbar based neuromorphic architectures

Python implementations and simulations of HP Labs Ion Drift and Yakopcic memristor models.

Dynex is a next-generation platform for neuromorphic computing based on a new flexible blockchain protocol. It is designed for the development of software applications and algorithms that utilize neuromorphic hardware and are capable of accelerating computation. To accomplish this goal, the platform connects hosts that are running clusters of neuromorphic chips with users and applications that utilize this next-generation hardware. On the TuringX platform, computation time is exchanged for the TuringX native token.

Brilliantly Radical Artificially Intelligent Neural Machine

With the end of Moore’s law approaching and Dennard scaling ending, the computing community is increasingly looking at new technologies to enable continued performance improvements. A neuromorphic computer is a nonvon Neumann computer whose structure and function are inspired by biology and physics. Today, such systems can be built and operated using existing technology, even at scale, and are capable of outperforming current quantum computers.

Dynex has also developed a proprietary circuit design, the Dynex Neuromorphic Chip, that complements the Dynex ecosystem and turns any modern field programmable gate array (FPGA) based chip into a neuromorphic computing chip that can perform orders of magnitude faster than classical or quantum methodologies for a wide range of applications. Due to the dominance of ASICs in the proof-of-work token mining industry, there is a large amount of dormant FPGA infrastructure available which can be converted into high performance next-generation neuromorphic computing clusters.