If your tool is a computer, everything begins to look like a list of numbers.
- Recap of the syntax of Python.
- How to start solving a problem using an algorithm.
- Thinking about algorithms in terms of how many operations they require and how much memory they use.
- Pancakes!
Data is the new bacon.
- How to organize a dataset.
- Thinking about instances and variables.
- Grouping data.
- Intuition about significance testing
Homework I out.
The eye is by far our most powerful tool of analysis.
- What are the different types of plots, and what can you use them for?
- Matplotlib essentials (= one hour of headaches).
- Tufte's information density theory of plotting.
- Plotting and cognition.
You find yourself enrolled in a program to become a master of matrices.
- How computers represent numbers.
- Basic operations on vectors and matrices.
Homework I due.
Why computers can (theoretically) learn anything.
- How is learning possible?
- Smoke detectors and the linear model.
- Implement the perceptron, a simple learning classification algorithm.
Homework II out.
Build a better perceptron
- Probabalistic models: Logistic regression.
- Energy-based learning, a general model.
Ninjas dress in black to hide in the background of your terminal window.
- Files and the file system.
- The philosophy of Unix.
- Writing Python scripts and executing them in the shell
Homework II due.
Bob Ross wants to remind you that there are no mistakes in painting, only happy accidents.
- Painting with numbers.
- Morphological operations.
Homework III out.
How to recognize trees from quite far away.
- Identifying and labeling parts of images.
- Going from continous coordinate spaces to pixel spaces.
Graphs are everywhere!
- Two strategies of unguided graph search: depth-first, breadth-first.
- Using heuristics to cut corners: A*-search.
- Graph search and finding your way on a map.
Homework III due.
- Graphs as Markov chains.
- Random walks, or what a drunken walk can find out.
- Analysis of social graph data sets: data from social networks.
Homework IV out.
How can Google find anything in no time?
- The inverted index data structure.
- Auto-completion using the Trie data structure.
Perceptrons are back, and they brought their friends.
- Feed-forward neural networks.
- Building blocks: layers, activation functions.
- Back-propagation.
Homework IV due.
- Clever hacks using neural networks: auto-encoders, word2vec.
Homework V out.