BME TMIT was officially nominated to NVidia GPU Education Center thank to successfully application from our department.
Graphic Processing Units (GPUs) are often used for scientific, massively parallel computations besides image and video rendering. Deep learning is such a field, which exploits the performance advantages of the GPUs. Lately deep learning has outperformed other machine learning algorithms in many research fields (including speech synthesis, speech recognition, image classification, natural language processing) and have already surpassed the accuracy of humans in some cases.
We are teaching GPU based deep learning in the Speech Technology and Smart Interactions Laboratory (SmartLab) of our department from 2014 in Project Laboratory 1-2, BSc Thesis Project and Diploma Thesis Design 1-2 courses. Thanks to the high interest a new course (Deep Learning in practice with Python and LUA) will help students to learn the latest deep learning algorithms, software and GPU architectures from September 2016.
In the framework of the NVidia GPU Education Center two Titan X, one Tesla K40 GPUs and books about GPU (CUDA) programming are donated for educational purposes. NVidia also gives access to additional course materials and hands-on lab exercises. The employees of BME can also get advices about hardware and software issues from NVidia experts.
‘I share the view of many researchers who claim that deep learning will become such a mainstream technology by 2020 as Internet is nowadays. It is very important that we keep up with the international progress and with the requirements of the industry and we teach the latest deep learning techniques, frameworks with state of the art GPU infrastructure. The NVidia grant is a superb opportunity in this progress!’ – said Dr. Bálint Pál TÓTH, the coordinator of the deep learning course.