machine-learning

Here are some ML examples implemented by python or octave during my learnings

Github projects & tools:

• Machine learning cheat sheet
• Stanford CoreNLP
• T2T: Tensor2Tensor Transformers
• SfMLearner
• Programming exercises for the Stanford Unsupervised Feature Learning and Deep Learning Tutorial
• An open-source software library for Machine Intelligence
• A Flexible and Efficient Library for Deep Learning
• scikit learn
• scikit learn videos

Other links:

• Andrew Ng's twitter & website
• Master machine learning by using it on real life applications
• Kaggle
• Sebastian Ruder's blog
• colah's blog
• What are some machine learning algorithms that you should always have a strong understanding of and why?

Youtube:

• Lecture 01 - The Learning Problem
• Deep Learning for Computer Vision (Andrej Karpathy, OpenAI)
• Nuts and Bolts of Applying Deep Learning

Courses:

• Machine Learning by Stanford University
• Machine Learning Foundations: A Case Study Approach by University of Washington
• Machine Learning: Regression by University of Washington
• Machine Learning: Classification by University of Washington
• Machine Learning: Clustering & Retrieval by University of Washington
• Deep Learning Specialization
• 15 hours of expert machine learning videos

From ML Mastery Blog:

Standard machine learning terms

There are key concepts in machine learning that lay the foundation for understanding the field.

I like to break them down into three areas:

• Data
• Learning
• Modeling

You need to know about data, such as a data instance, data feature and datasets like test and training datasets.

Learning is a big topic. Induction is the main type of learning used by machine learning algorithms (this is generalizing from specific examples). With generalizing there is the risk of over-learning (too specific) and under-learning (too general).

Then there is the type of learning, such as online or offline (real-time or batch) and supervised or unsupervised (teacher or no teacher).

The model is the artifact created from running an algorithm. It is the "thing" that you can use to describe the structure in the data or make predictions from new data.

There are two big issues when it comes to the model and that is the bias (the assumptions made to create the model) and the variance (how sensitive it is to data). You may have heard of the bias-variance trade-off that seeks a balance between these two concerns.

Learn more about the standard machine learning terms in this post