AI Saturdays Lesson Plan 

Here are learning roadmaps organised in weeks. Each week consists of lecture videos, 
reading materials and programming assignments. Feel free to head to the A
​ I6 forums​ to 
discuss or ask questions you might have. 
 

Machine Learning from Scratch  

This is for anyone who wants to learn Machine Learning or TensorFlow. The only 
prerequisite is high school math and a ton of passion. We will follow ​Machine Learning 
Crash Course​ by Google - it breaks down concepts of Machine Learning in fun, bite-sized 
portions and provides thoughtful exercises to deepen your understanding.   
 
Timeline  Learning material   Assignments  

Week 1/  Prerequisites: Linear Algebra  Assignments and 

Week of 4    readings 
August  - Khan academy Linear Algebra​ (​Finish Complete 
2018  Vectors and Spaces and Matrix Transformations 
modules) 
 
- Still hungry for more? Try ​Essence of linear 
algebra 

Week 2/  Prerequisites: Python basics  Finish 5 python 

Week of 11    exercises ​here 
August  For those ​without​ any programming experience:  (solutions provided) 
2018  - Automate the boring stuff with python: 
Chapter 0 
- Python for Data Science​ (7 hours) 
 
For those ​with​ programming experience: 
- Automate the boring stuff with python: 
Chapters 1 - 5 

Week 3/  Prerequisites: Python advanced  Replicate the codes in 

Week of 18  “A Byte of Python” 
August   
2018  - A Byte of Python ebook: O​ bject oriented  Create a Cheat Sheet 
programming   summarising all that 
- Udemy Numpy stack in Python​: Numpy,  you have learnt about 
Matplotlib, Scipy, Pandas  Python.   
- MIT Course Object Oriented Programming 

Week 4/  Recommended reading:   Complete “Check Your 

Week of 25  - Machine Learning is Fun!  Understanding”  
August     
2018    Discuss: Artificial 
Machine Learning Crash Course by Google:    Intelligence, Machine 
  Learning, Deep 
- Introduction to ML  Learning, Data Science 
   
- Framing    
 
- Descending into ML  
 
 

Week 5/  Machine Learning Crash Course by Google​:  Complete “Check Your 
Week of 1    Understanding”  
September  - Reducing Loss   
2018  Code up a simple 
gradient descent 
algorithm on a linear 
regression problem 
(​example​). 
 
Linear regression 
from scratch 
 
Another Linear 
regression from 
scratch in Python  

Week 6/  This week, explore different frameworks for  Programming exercise 
Week of 8  Machine Learning. Recommended readings:  with tensor flow  
September  - What are Machine Learning Frameworks   
2018    Official scikit-learn 
- Best Python libraries for Machine Learning  tutorial  
and Data Science 
 
- TensorFlow or Keras? 
 
- Introduction to scikit-learn 
 
- Machine Learning Crash Course by Google​: 
First Steps with TensorFlow 
  
Note: Don’t spend too much time on picking a framework. If you 
cannot decide, try out TensorFlow first, since assignments 
offered in the Crash Course are in TensorFlow. 
Week 7/  Machine Learning Crash Course by Google​:  Complete “Check Your 
Week of 15  - Generalization  Understanding” and 
September    “Programming 
2018  - Training and Test Sets   Exercises”, if any  
Validation 
 
- Representation   

Week 8/  Machine Learning Crash Course by Google   Complete “Check Your 
Week of 22    Understanding” and 
September  - Feature Crosses  “Programming 
2018  Exercises”, if any  
 

Week 9/  Machine Learning Crash Course by Google   Complete “Check Your 
Week of 29    Understanding” and 
September  - Regularization: Simplicity   “Programming 
2018    Exercises”, if any 
- Logistic Regression    
Logistic Regression 
with Tensor Flow 
 
Logistic Regression 
from scratch in 
Python 
 
Titanic challenge using 
sklearn 
 
Alternatively, find your 
own dataset to 
perform logistic 
regression 
 

Week 10/  You are at the halfway mark! Use this week to take a good break and re- 
Week of 6  energize.  
October 
2018 

Week 11/  Machine Learning Crash Course by Google   Complete “Check Your 
Week of 13    Understanding” and 
October  - Classification  “Programming 
2018    Exercises”, if any 
- Regularization: Sparsity    
AUC curve:​ discussion 
and analogy 
  
Evaluating a 
classification model 
(using scikit learn) 
 
K-means clustering 
algorithm​ (hands-on 
assignment) 

Week 12/    Complete “Check Your 

Week of 20  Video Lecture: ​What is a Neural Network?  Understanding” and 
October    “Programming 
2018  Machine Learning Crash Course by Google   Exercises”, if any 
   
- Introduction to Neural Nets  Create a Neural 
  Network using Keras 
 
Implementing a 
Neural Network from 
Scratch in Python 

Week 13/  Video Lectures:  Complete “Check Your 

Week of 27  - Gradient descent   Understanding” and 
October    “Programming 
2018  - What is backpropagation really doing?  Exercises”, if any 
   
Machine Learning Crash Course by Google   Tutorial on 
  Backpropagation 
- Training Neural Nets  (highly recommended) 
 
- Multi-Class Neural Nets 
 

Week 14/  Machine Learning Crash Course by Google   Complete “Check Your 
Week of 3    Understanding” and 
November  - Embeddings   “Programming 
2018    Exercises”, if any 
- All sections under “ML Engineering” and “ML 
Real World Examples”  
 

Deep Learning Specialization Lesson Plan  

This lesson plan is for those who have understood the basics of Machine Learning and 
would like to gain a deeper understanding in Deep Learning. It follows ​deeplearning.ai​’s 
Deep Learning Specialization​ on Coursera. This course is one of the ​highest rated​ courses 
on Deep Learning.  
 
 
Neural Networks and Deep Learning (Weeks 1-6) 
Period    Youtube lecture videos​ (each video is  Assignment  
labelled with an alphanumeric string, 
e.g C
​ 1W1L01) 

Week 1/ 4  Revision on Python  

August 
2018 

Week 2/  C1W1L01 - C1W1L06  Assignment 1 

11 August 
2018 

Week 3/  C1W2L01 - C1W2L06  Assignment 2 

18 August 
2018 

Week 4/  C1W2L07 - C1W2L18  Assignment 3 

25 August 
2018 

Week 5/  C1W3L1 - C1W3L11  Assignment 4 

1 
Septembe
r 2018 

Week 6/  C1W4L01 - C1W4L18  Assignment 5 

8 
Septembe
r 2018 

Week 7/  Break  

15 
Septembe
r 2018 
 
 
Improving deep neural networks: hyperparameter (Weeks 8-13) 
 
Period    Youtube lecture videos​ (each video is  Assignment  
labelled with an alphanumeric string, 
e.g C
​ 2W1L01) 
 Week 8/  C2W1L01 - C2W1L06  Assignment 1  
22 
Septembe
r 2018  

Week 9/  C2W1L07 - C2W1L14  Assignment 2 

29 
Septembe
r 2018  

Week 10/  C2W2L01 - C2W2L05  Assignment 3 

6 October 
2018 

Week 11/   C2W2L06 - C2W2L09  Assignment 4 

13 
October 
2018 

Week 12/   C2W3L01 - C2W3L05  Assignment 5 

20 
October 
2018 

Week 13/   C2W3L06 - C2W3L11  Assignment 6 

27  NOTE: C2W3L10 and C2W3L11 are 
October  mislabeled as C2W2L10 and C2W2L11 
2018  respectively. 
 
 
Research-focused Lesson Plan 
Learners of this track are those who have acquired the prerequisites to understand and 
implement state of the art AI models. They can choose from one or more of the following 
courses : UCL course on Reinforcement Learning (UCL RL), Berkeley CS294, Stanford 
Natural Language Processing (CS224n) and Stanford Computer Vision (CS231n).  
 
These courses are offered by their respective universities and (fortunately for us) are freely 
accessible by the public. Learning materials for the courses can be quite ‘scattered’, i.e 
lecture videos for these courses are not found on the official website but on youtube; they 
are also not aligned with the course outline on the official website. We will follow one 
lecture video from youtube every week and refer to the official pages for lecture notes or 
assignments. 
 
Important links  
UCL RL  CS231n (Computer vision)  CS224n (NLP) 
 ● Lecture Videos​ from  ● Lecture Videos​ from  ● Lecture Videos​ from 
youtube  youtube  youtube 
● Official Page   ● Official Page  ● Official Page  
● Code and exercises  ● Assignments  ● Lecture Notes​ (click on 
● Berkeley CS294  ‘notes1.pdf’ or similar) 
(advanced RL)  ● Independent Review of this 
course 
 
 
Lesson schedule 
 
Week of   UCL RL  CS231n (Computer  CS224n (NLP) 
vision) 

4 August  Revision of prerequisites:  See “​Module 0​”  See “​Suggested readings 

2018  Python, Linear Algebra etc    on Jan 9​” 
    Assignments 

11 August  UCL/Deep Mind  Youtube lecture video:  Youtube lecture video: 

2018  Reinforcement Learning  Image Classification  Word Vector 
Lecture 1 – Intro to    Representations:word2v
Reinforcement Learning  Course Notes  ec 
     
Assignments/ Links to  Assignments  Assignments and 
materials  readings 
 

18 August  UCL/Deep Mind  Youtube lecture video:  Youtube lecture video: 

2018  Reinforcement Learning  Loss Functions and  Global Vectors for Word 
Lecture 2 – Markov Decision  Optimization  Representation 
Processes     
  Assignments and readings  Lecture Notes 
Assignments/ Links to   
materials  Assignments and 
readings 

25 August  UCL/Deep Mind  Youtube lecture video:  Youtube lecture video: 

2018  Reinforcement Learning  Introduction to Neural  Word Window 
Lecture 3 – Planning by  Networks   Classification and Neural 
Dynamic Programming    Networks 
  Assignment and readings    
Assignments/ Links to  Lecture Notes 
materials 

1  UCL/Deep Mind  Youtube lecture video:  Youtube lecture video: 

Septembe Reinforcement Learning  Convolutional Neural  Backpropogation 
r 2018  Lecture 4 – Model-free  Networks  Assignments and 
Prediction    readings 
  Assignments and readings 
 Assignments/ Links to 
materials 

8  UCL/Deep Mind  Youtube lecture video:  Youtube lecture video: 

Septembe Reinforcement Learning  Training Neural Networks  Dependency parsing 
r 2018  Lecture 5 – Model-free  Part 1   
Control    Lecture Notes 
   
Assignments and 
readings 

15  Break  
Septembe
r 2018 

22  UCL/Deep Mind  Youtube lecture video:  Youtube lecture video: 

Septembe Reinforcement Learning  Training Neural Networks  Introduction to 
r 2018   Lecture 6 – Value Function  Part 2  TensorFlow (optional) 
Approximation     
  Youtube lecture video: 
Recurrent Neural 
Networks and Language 
Models 
 
Lecture Notes 
 
Assignments and 
readings 

29  UCL/Deep Mind  Youtube lecture video:  Youtube lecture video: 

Septembe Reinforcement Learning  Deep Learning Software  Machine Translation and 
r 2018   Lecture 7 – Policy Gradient  Advanced Recurrent 
Methods  LSTMs and GRUs 
   
   
 
 

6 October  UCL/Deep Mind  Youtube lecture video:  Youtube lecture video: 

2018  Reinforcement Learning  CNN Architectures  Neural Machine 
Lecture 8 – Integrating    Translation and Models 
Learning & Planning  with Attention 
   

13  UCL/Deep Mind  Youtube lecture video:  Youtube lecture video: 

October  Reinforcement Learning  Recurrent Neural  Gated Recurrent Units 
2018  Lecture 9 – Exploration &  Networks  and Further Topics in 
Exploitation  NMT 
 

20  UCL/Deep Mind  Youtube lecture video:  Youtube lecture video: 

October  Reinforcement Learning  Detection and  End-to-End Models for 
2018  Lecture 10 – Classic Games  Segmentation  Speech Processing  
  

27  Break  
October 
2018 

3  Berkeley CS294 Deep  Youtube lecture video:  Youtube lecture video: 

November  Reinforcement Learning  Visualizing and  Tree Recursive Neural 
2018    Understanding  Networks and 
  Constituency Parsing 

10  Berkeley CS294 Deep  Youtube lecture video:  Youtube lecture video: 
November  Reinforcement Learning  Generative Models  Coreference Resolution 
2018     

17  Berkeley CS294 Deep  Youtube lecture video:  Youtube lecture video: 
November  Reinforcement Learning  Deep Reinforcement  Dynamic Neural 
2018    Learning  Networks for Question 
  Answering 

24  Berkeley CS294 Deep  Youtube lecture video:  Youtube lecture video: 
November  Reinforcement Learning  Efficient Methods and  Issues in NLP and 
2018  Hardware for Deep  Possible Architectures in 
Learning  NLP 
 

1    Youtube lecture video:  Youtube lecture video: 

December  Berkeley CS294 Deep  Adversarial Examples and  Tackling the Limits of 
2018  Reinforcement Learning  Adversarial Training  Deep Learning for NLP 
   

Until  Work on projects  

January 
2019 
 
Bonus tips  
- The lecture videos go up to 2 hours; watch the lecture video at 2x speed, slowing 
down only at parts that cover important concepts 
- Revise and make sure that you have ​mastered t​ he fundamentals of deep learning  
- Don’t aim to understand everything and anything.