Description
The software library lib4neuro is being developed at IT4Innovations national supercomputing center, VSB - Technical University of Ostrava. The main purpose of this library is the construction of artificial neural networks and their subsequent utilization in the field of physical chemistry.
Currently implemented training methods:
- Steepest descent method
- Barzilai-Borwein method
- Mini-batch gradient descent method
- Levenberg-Marquardt method
- Globalized particle swarm optimization
The code is written in C++17 and can be compiled with CMake 3.12.* and higher, both on Linux and Windows systems. lib4neuro is using the following external libraries:
If those libraries are not installed in the system already, lib4neuro is able to download and compile them locally.
Authors
- Martin Beseda (
This email address is being protected from spambots. You need JavaScript enabled to view it. ) - Martin Mrovec (
This email address is being protected from spambots. You need JavaScript enabled to view it. ) - Michal Kravčenko (
This email address is being protected from spambots. You need JavaScript enabled to view it. ) - David Vojtek (
This email address is being protected from spambots. You need JavaScript enabled to view it. )
Download
The library can be downloaded from our public repository https://code.it4i.cz/moldyn/lib4neuro.
Acknowledgements
This software has been implemented with the financial support of:
- The Ministry of Education, Youth and Sports from the National Programme of Sustainability (NPS II) project „IT4Innovations excellence in science - LQ1602“
- The project PHC BARRANDE 2018 (40688TH)
- The project SGS Internationalization of the doctoral studies in the area of molecular physics (SP2018/178)