Course information

1. Course logistic. Machine Learning, Deep learning. Computer vision.

2. Image classification(what is nnets) + needfull tools for this course (python(pycharm), numpy, pytorch, tutorial for gcloud).

3. How to train: loss function and optimization, backpropogation.

4. How to train: data representation (image), batch normalization, dropout.

5. What is convolution + pytorch.

6. Pytorch (tensor, datasets, cuda, nets on pytorch, cuda) + pytorch tutorials.

7. Into kaggle + hardware.

8. The main architectures.

9. Visualizing and Understanding (Feature visualization and inversion Adversarial examples DeepDream and style transfer).

10. Transfer learning + tips and tricks (augmentation, ensembles + cross-validation, mixup, labelsmoothing + lr_schedulers).

11. Another comuter vision problems (segmentation, detection).

12. The most popular tasks today (face recognition, self-diving, deepfake, gans).

13. May be some overview of articles.

14. Optimization nnets (tensorrt, pruning, knowledge distillation).

Educational designer:

Georgy Surin, computer vision / deep learning engineer at Inspector-cloud

suringeorgy1@gmail.com

kaggle.com/formemorte

3 silver medal, 1 bronze

7th place at Seismic challenge (https://boosters.pro)

Teaching assistants