Aidge#

is an open-source deep learning platform specialized in the design of deep neural networks intended to operate in systems constrained by power consumption or dissipation, latency, form factor (dimensions, size, etc.), and/or cost criteria.

Aidge’s eventual ambition is to offer:

an integrated approach encompassing the entire design flow, from application development to deployment: data formatting, neural network exploration, learning, testing, and optimized code generation;
several functions to reduce the computational complexity of models and their memory requirements, most often using quantization (during or after training) and topological optimization techniques;
compatibility with a wide range of commercially available hardware targets, offering optimized implementations for MCUs and DSPs, GPUs, FPGAs or NPUs;
a modular design and simple abstraction layer, so features can be added and modified with ease, including the low-level implementation of calculation functions depending on the specific characteristics of the hardware being deployed (approximate calculation, specific saturated arithmetic, etc.);
a high degree of interoperability, with support for the ONNX standard and integration with the PyTorch and Keras platforms;
a multiparadigm approach that integrates the simulation of neuromorphic neural network models into the same platform;
sovereignty and control of the code, as Aidge is independent of other deep learning platforms.

For more detailed technical insights into some interesting Aidge’s features, please have a look to Why Aidge?.

License#

Aidge is released under the Eclipse Public License 2.0