Analytics to Reinforcement Learning 5 weekends course
This is a 5 weekends course covering:
Intro to Data Analytics, Oct 21-22
Intro to Machine Learning, Nov 4-5
Intro to Deep Learning, Nov 18-19
Advanced Deep Learning, Dec 2-3
Reinforcement Learning, Dec 16-17
The course provides a comprehensive introduction to data science with deep dives in data analytics, machine learning, deep learning and reinforcement learning. It is meant to provide a solid base to build deeper knowledge in the field.
First weekend:Intro to Data Analytics with Python, SQL, Spark and Seaborn
Use Python and Pandas to select, group and summarize your data
Decide what data to keep and what to ignore
Create compelling visualizations using Seaborn and Matplotlib
Connect and retrieve data from a database using Python
Extend your analyses to relational databases using SQL
Perform aggregations and combinations using SQL
Include unstructured data sources in your analysis using Spark
Scale up your analyses to Gygabytes of data using Spark on AWS
Combine Spark and SQL for maximum flexibility and power
Second weekend:Intro to Machine Learning with Python & Scikit-Learn
Recognize problems that can be solved with Machine Learning
Select the right technique (is it a classification problem? a regression? needs preprocessing?)
Load and manipulate data with Pandas
Visualize and explore data with Matplotlib and Bokeh
Build regression, classification and clustering models with Scikit-Learn
Evaluate model performance with Scikit-Learn
Build, train and serve a predictive model using Python, Flask and Heroku
Third weekend:Intro to Deep Learning with Python (Keras/Tensorflow)
Fundamentals of deep learning theory
How to approach and solve a problem with deep learning
Build and train a deep fully connected model with Keras
Build and train a Convolutional Neural Net with Keras on a cloud GPU machine
Build and train a Recurrent Neural Net with Keras on a cloud GPU machine
Application to Image processing/Text processing
Fourth weekend:Advanced Deep Learning with Python & Tensorflow
Review of fundamental deep learning architectures (Fully Connected, Convolutional, Recurrent)
Build and train a model with pure Tensorflow
Online training / continous training
Custom architectures and loss functions
Review of famous architectures (Inception, Wavenet)
Setting up a machine for deep learning / serving a model
Fifth weekend:Reinforcement Learning with Python, Tensorflow and OpenAI
Train neural networks to play video games using Deep Q-Learning
Reduce the dimensionality of your data using autoencoders
Improve the efficiency of your algorithms with generative adversarial networks
Train AI agents to interact in an environment using OpenAI Gym and Universe
Train a Word2Vec model to encode natural language
Is lunch provided
Yes! Lunch is included.
Are there any prerequisites?
Previous experience programming in Python or in other languages is advised to make best use of the workshop.
In the last 2 years Python has become a de-facto standard in data science and is widely adopted by most major companies. Reasons for this success include:
large set of mature data science libraries => most needs covered
worldwide community of enthusiasts => get help when you need it
easy to learn, read and write => start contributing immediately
supports both functional and object oriented coding => versatile and powerful
full stack programming language => easier interaction between data scientists and software engineers
SQL is the most widely used language for managing data in a relational database. It is supported by both open source projects like MySQL and PostgreSQL and by enterprise databases like Oracle, Microsoft SQL Server and many others.
Apache Spark has revolutionized how we build and deploy data pipelines for ETL, Visualization and Machine Learning. Reasons for this success include:
Flexible enough to run SQL-style queries, machine learning algorithms, and everything in between
Fast and scalable: efficient memory use => runs up to 100x faster than Hadoop
Supports data exploration and production workflows => same code that works on a laptop can be deployed to cloud-based computing clusters
Free and open-source
Keras is a high-level neural networks api and library that allows to simply build and train deep learning models using Tensorflow or Theano as backend. Written in Python it focuses on enabling fast experimentation. It recently became the preferred high level api for Tensorflow and it thus provides a great entry point to approach Tensorflow. Keras highlights:
Allows for easy and fast prototyping
Supports Fully connected, Convolutional and Recurrent
Supports arbitrary connectivity schemes
Runs seamlessly on CPU and GPU
Integrates very well with Tensorflow and Tensorboard
There are many open source Deep Learning libraries. Tensorflow is backed by Google and is quickly becoming one of the most used libraries in the fields. It has a large and growing community of users and it is versatile and easy to learn. Highlights include
largest community of developers
state of the art models and nodes
high scalability, can be distributed on many GPUs
production performance and deployment tools
very versatile and powerful for distributed high performance computing beyond neural networks
The course is lead by Francesco Mosconi. Ph.D. in Physics and Data Scientist at Catalit LLC, he was formerly co-founder and Chief Data Officer at Spire, a YC-backed company that invented the first consumer wearable device capable of continuously tracking respiration and physical activity. Machine Learning and python expert he also served as Data Science lead instructor at General Assembly and The Data incubator.
Read our reviews on Yelp
In certain cases, we may need to cancel this workshop due to circumstances beyond our control or otherwise. If this happens, we will refund all registration fees for those who signed up. We are not responsible for any related expenses incurred by registered attendees (including but not limited to travel and hotel expenses).
More than 1 week before course: full refund.
Less than 1 week before course: no refund available.
All public workshops come with a no-questions-asked money-back guarantee. If you are unhappy for any reason after attending the class, you can ask for a full refund.