Cognitive Class

Accelerating Deep Learning with GPU

Training complex deep learning models with large datasets take long time. In this course you learn how to use accelerated hardware to overcome the scalability problem in deep learning.

Start the Free Course

In the previous courses, we learned to use TensorFlow and Deep Leaning, providing some simple and fast-to-run examples. But, have you ever tried to train a complex deep learning models with huge data?  You should expect hours, days or weeks sometimes to train a complex model with large dataset. So, what is the solution?

Well, you should use accelerated hardware, for example, you can use Google’s Tensor Processing Unit (TPU) or Nvidia GPU to accelerate your convolutional neural network computations time on the could. These chips are particularly designed to support the training of neural networks, as well as the use of trained networks (inference). These accelerating hardwares have recently succeed to reduce the training time several times.

But the problem is that your data might be sensitive and you may not feel comfortable to upload it into public cloud, and you need to analyze it on-premise.  In this case, you need to use an in-house system with GPU support. One solution is using IBM’s Power Systems with Nvidia GPU, and PowerAIBuilt on IBM’s Power Systems, PowerAI is a scalable software platform that accelerates deep learning and AI with blazing performance for individual users or enterprises. The PowerAI platform supports popular machine learning libraries and dependencies including Tensorflow, Caffe, Torch, and Theano. 


Please note that this course is a weekly scheduled course with limited enrollment. We will only accept the first 20 people who enroll to this course. The course will be started every Monday at 4am UTC and it will be closed after one week. Please note that you are only allowed one enrollment and one attempt at the content for this course. Enrollment will be on a First Come, First Served basis.


Course Syllabus

Module 1 –  Quick review on Deep Learning

  • Intro to Deep Learning
  • Deep Learning Pipeline

Module 2 –  Hardware Accelerated Deep Learning

  • How to accelerate a deep learning model?
  • Running TensorFlow operations on CPUs vs. GPUs
  • Convolutional Neural Networks on GPUs
  • Recurrent  Neural Networks on GPUs

Module 3 – Deep Learning in the Cloud

  • Deep Learning in the Cloud
  • How does one use a GPU
  • Stock Price Prediction

Module 4 – Distributed Deep Learning

  • Distributed Deep Learning

General Information

  • This course is free.
  • This course if with Python language.
  • It is self-paced.
  • It can be taken at any time.
  • It can be audited as many times as you wish.

Recommended skills prior to taking this course

  • Neural Network


  • Python programming
  • Tensorflow
  • Deep Learning fundamental

Course Staff

Dr. Saeed Aghabozorgi, TensorFlow Course Instructor

Saeed Aghabozorgi, PhD is a Sr. Data Scientist in IBM with a track record of developing enterprise level applications that substantially increases clients’ ability to turn data into actionable knowledge. He is a researcher in data mining field and expert in developing advanced analytic methods like deep learning, machine learning and statistical modelling on large datasets.