Zero to Hero in Machine Learning with Keras
Download as PPTX, PDF
4 likes1,260 views
Presentation for Deep Learning NYC Meetup. Explaining how machine learning works, relevant terminology, and how to get started in it with Python and Keras.
Zero to Hero in Machine Learning with Keras
- 1. Zero to Hero in Machine Learning with Keras Boris Yakubchik (Forbes) June 7, 2018 Deep Learning NYC
- 2. Outline 1. About ML 2. How to get started 3. Workflow 4. Tools you can use 5. Examples 6. Q&A
- 3. Deep Learning with Python Fran巽ois Chollet
- 4. What is ML?
- 6. Types of learning Supervised Unsupervised Self-supervised Reinforcement
- 10. Vocabulary Layers Tensors Gradient descent Optimizer Loss Activation Overfitting Back propagation Epoch
- 11. Gradient descent
- 13. Loss
- 14. Activation Function ReLU (Rectified Linear Unit) Sigmoid
- 15. Overfitting
- 16. Overfitting
- 17. Vocabulary Layers Tensors Gradient descent Optimizer Loss Activation Overfitting Back propagation Epoch
- 18. Architectures Dense Neural Network Convolutional Neural Network (CNN or ConvNet) Recurrent Neural Network (RNN)
- 19. Dense
- 20. ConvNet
- 21. ConvNet
- 23. ConvNet
- 24. RNN
- 25. Getting started 1. Install python 3.6+ 2. Install pip (Pip Installs Python) 3. Install venv (virtual environment) 4. alias py="python3 cd my-1337-ml-dir py -m venv venv source venv/bin/activate venv/Scripts/activate.bat if on PC pip install -r requirements.txt
- 26. keras tensorflow # get that GPU speed! tensorflow-gpu numpy==1.13.0 # tools jupyter # html5lib newer version needed to get tensorboard to work html5lib==1.0.1 # data management pandas # utils pylint autopep8 matplotlib # required for PIL -- image manipulation sometimes pillow opencv-python requirements.txt
- 27. Workflow 1. Define problem 2. Assemble dataset 3. Choose measure of success 4. Decide evaluation protocol 5. Prepare your data 6. Develop a model 7. Iterate models
- 28. Tools
- 29. TensorFlow
- 30. Keras =
- 31. OpenRefine
- 32. Jupyter Notebook
- 33. Jupyter Lab
- 34. TensorBoard
- 35. DEMO