Applied AI from Scratch in Python Training Course

Supervised learning: classification and regression

• Machine Learning in Python: intro to the scikit-learn API
  • linear and logistic regression
  • support vector machine
  • neural networks
  • random forest
• Setting up an end-to-end supervised learning pipeline using scikit-learn
  • working with data files
  • imputation of missing values
  • handling categorical variables
  • visualizing data

Python frameworks for for AI applications:

• TensorFlow, Theano, Caffe and Keras
• AI at scale with Apache Spark: Mlib

Advanced neural network architectures

• convolutional neural networks for image analysis
• recurrent neural networks for time-structured data
• the long short-term memory cell

Unsupervised learning: clustering, anomaly detection

• implementing principal component analysis with scikit-learn
• implementing autoencoders in Keras

Practical examples of problems that AI can solve (hands-on exercises using Jupyter notebooks), e.g. 

• image analysis
• forecasting complex financial series, such as stock prices,
• complex pattern recognition
• natural language processing
• recommender systems

Understand limitations of AI methods: modes of failure, costs and common difficulties

• overfitting
• bias/variance trade-off
• biases in observational data
• neural network poisoning

Applied Project work (optional)