Duration: 2-3 hours
Training type: classroom or online
Rather than lecturing, with Panda Training Jerome sees his role as a coach, helping you in your path to AI. The courses he offers are designed to give both background information and hands-on experience. During a 2-3 hours session you will get insights about a specific topic, and after that you will work on a 20 hour project. While doing the project, you will have a dedicated Slack channel to discuss ideas and get support. At the end your implementation will be reviewed and you will get feedback.
- Getting familiar with deep neural networks, background and glossary.
- Presentation of different tasks or applications (vision, NLP, translation, Deep RL..)
- Presentation of the different architectures (CNNs, RNNs, LSTMs..)
- Supervised vs unsupervised techniques
- Presentation of the different building blocks (feed forward, convolution, regularization)
- Different frameworks and architectures
- Debugging and visualization of Neural networks
Homework: Implement an image classifier for MNIST, measure, visualize layers, proposed framework: Keras (Tensorflow).