notes-shivam

AI-ML

4 min read

My plan is to present these information in a more professional manner, i.e. on YouTube. But I just saw your post and I thought ‘why not post my notes for this video I’m planning on this matter?’ So here you go, consider this a sneak peak ;)

Level 1 – Informed Decision Maker:

  • Basic understanding of what ML is and what it is not

  • Know how ML can/will affect their lives in the short to mid term

  • Know how ML‘s potential can be utilized to serve themselves (or their teams)

resources:

coursera – ai for everyone andrew ng – machine learning yearning coursera – machine learning (first three weeks) 100 page ML book

From now on, three areas of focus will be given for each level: Mathematics, Concrete ML knowledge, and Programming

Level 2 – Competent Developer

  • Have basic intuition about the math relevant for ML

  • Know the theory behind the most basic ML algorithms (linear/logistic regression, svm, decision trees/random forests, knn clustering, basic neural networks)

  • Know the basics of the Python programming language, the data science stack (numpy, pandas, matplotlib/seaborn, sklearn, sql queries) and how to implement basic ml pipelines

Mathematics:

Linear Algebra: - Gilbert Strang – MIT online lecture (find problems and solutions) - 3blue1brown – Essence of linear algebra

(Multivariate) Calculus: - 3blue1brown – Essence of calculus - Khan Academy – AP/College Calculus AB - Khan Academy – Multivariable calculus

Statistics and Probability: - Khan Academy – Statistics and Probability

Concrete ML Knowledge:

coursera – machine learning coursera – deep learning specialization (courses 1 to 4 on youtube) Dmitry Kobak – introduction to machine learning

Programming:

Corey Schafer – Python Programming Beginner Tutorials Corey Schafer – Python OOP Tutorials – Working with Classes Corey Schafer – Jupyter Notebook Tutorial Corey Schafer – Pandas Tutorials Corey Schafer – Matplotlib Kaggle – Microcourses Keith Galli - Complete Python NumPy Tutorial Streamlit data science handbook (a bit verbose for self study) WQU - Applied Data Science Module: - Applied Data Science I: Scientific Computing & Python - Applied Data Science II: Machine Learning & Statistical Analysis

Level 3 – Expert Developer

  • Have enough mathematical proficiency to be able to read academic papers or graduate level textbooks about ML

  • Have extensive knowledge and understanding of a wide range of ML algorithms to be able to apply the correct the algorithm for the problem at hand. Be able to discuss the pros and cons of different algorithms and consult decision makers

  • Know how to address the challenges of dealing with stochastic code and be able to create complex ml pipelines that can be integrated into larger software infrastructures

Mathematics:

Statistics and Probability: - MITx – Probability-The Science of Uncertainty and Data - MITx – Fundamentals of Statistics

Wide range of topics: - Ulrike von Luxburg – Mathematics of Machine Learning

Concrete ML Knowledge:

  • Kilian Weinberger: Machine Learning for intelligent Systems
  • Andreas Geiger: Deep Learning
  • Ulrike von Luxburg: Statistical Machine Learning

Programming:

  • hands-on machine learning with scikit-learn, keras and tensorflow
  • Jose Portilla (Udemy): Python for Computer Vision with OpenCV and Deep Learning
  • Jose Portilla (Udemy): NLP - Natural Language Processing with Python
  • fast.ai
  • d2l
  • Soledad Galli:
    • deployment of machine learning models,
    • feature engineering for machine learning
    • feature selection for machine learning
    • testing and monitoring machine learning model deployments
    • machine learning with imbalanced data
  • Refactoring Guru Design Patterns
  • udacity courses

Level 4 – PhD level

  • Deepen understanding of advanced mathematics and selected branches of ML to be able to read exotic/very theoretical papers, perhaps even contribute by creating theoretical insights on your own

  • Be able to contribute to open-source projects and create innovative software products yourself

The resources for this level are more free-form, depending on your specialization:

textbooks

papers with code fast.ai 2 fast.ai code first introduction to nlp fast.ai numerical linear algebra

AMMI - Geometric Deep Learning Course steve burton - machine learning and dynamical systems tübingen – probabilistic machine learning tübingen – computer vision penn university – graph neural networks stanford – reinforcement learning deepmind – introduction to reinforcement learning stanford – natural language processing with deep learning openmined – private ai series

machine learning street talk lex friedman (not terribly rigorous but inspiring for finding your own directions of focus)

(And of course my YouTube channel: https://www.youtube.com/channel/UCg5yxN5N4Yup9dP_uN69vEQ)

Graph View