More Machine Learning From Udacity
Written by Sue Gee   
Friday, 20 November 2015

Udacity has opened a new Nanodegree in Machine Learning. At advanced level it prepares you to apply predictive models to massive data sets in fields like education, finance, healthcare or robotics.



The curriculum for the Machine Learning Engineer Nanodegree  has been developed with Google and in his email announcement Sebastian Thrun played up this selling point: 

If you followed the news, you saw that our partner, Google, just open-sourced TensorFlow, its deep learning software. And get this: our program already has a class on deep learning in TensorFlow in development, overseen by Vincent Vanhoucke, technical lead and manager in Google's deep learning infrastructure team.

Over the 10 to 12-month period required to complete the project-based certification students will:


  • Analyze the class and complexity of a given problem, and identify an appropriate algorithm and/or tools to apply towards solving it (e.g. gesture recognition, robot control).

  • Design an intelligent system that can act on the basis of input data towards optimizing some desired goal metric, with minimal supervision from a human.

  • Analyze the performance of an intelligent algorithm / system and present key metrics (such as accuracy, recall, computing time, etc. as appropriate) in an easy-to-understand and visually appealing form Handle the entire machine learning pipeline, from data to system.

The four set projects and the topics covered by their supporting courses are:

Project 1: Predicting Boston Housing Prices
Statistical techniques, Model Evaluation, Validation and Optimization.

Project 2: Building a Student Intervention System
Supervised Learning, Decision Trees, Regression, Neural Networks, Kernel Methods, Instance Based Learning, Bayesian Learning, Ensemble Learning.

Project 3: Creating Customer Segments
Clustering, Feature Scaling, Feature Selection, Feature Transformation, Semi-Supervised Learning.

Project 4: Train a Smartcab How to Drive
Markov Decision Processes, Reinforcement Learning, Game Theory.

The fifth and final component is a Capstone Project in which students apply multiple learning techniques to design an application to solve a problem of their own choice.

Enrollment in a Nanodegree program is subscription based at a fee of $200 per month after a 1-week free trial. If you graduate in less than 12 months you get half your tuition fee back. 



For those who are interested in the course content without wanting a certificate or having their projects evaluated, a lot of the materials are already available online for free. The links provided are:




Specific to Project 1:



Specific to Project 4:


Not only is Machine Learning a hot topic, it is one that seems to fit very well with the MOOC model and this is likely to be a popular Nanodegree.


The University of Tübingen's Self-Driving Cars Course

The recorded lectures and the written material of a course on Self-Driving Cars at the University of Tübingen have been made available for free. It's a first class opportunity to learn the in an [ ... ]

Supersimple - Deep Insights From Data

Announcing $2.2 Million in pre-seed funding, the Estonian startup Supersimple has launched an AI-native data analytics platform which combines a semantic data modeling layer with the ability to answer [ ... ]

More News


raspberry pi books



or email your comment to:





Last Updated ( Friday, 20 November 2015 )