100 Days Of ML Code — Day 028

100 Days Of ML Code — Day 028

Recap from Day 027

In day 027, we continued with how machine learning in the arts is different and brought it to an end. We saw that for most artistic applications, you are an expert on what the model you’re building should ultimately do. You will know whether your gesturally-controlled instrument is making a sound you’re happy with. You will know whether your dance motion classifier is accurately labeling the different actions a dancer is taking. This means that you’re not restricted to conventional metrics like cross-validation when you want to understand whether the model you’ve built is any good.

Today, we’ll start looking at generating useful inputs for machine learning.

Generating Useful Inputs For Machine Learning

We’ve now seen quite a variety of supervised and unsupervised algorithms for both classification and regression. If you’ve been following the article from the day 001, you should have a pretty good picture in your head of each training example living in a feature space, an idea of how different algorithms use examples in feature space to build models.

[Source](https://cdn.hashnode.com/res/hashnode/image/upload/v1632827282526/weyMyOwur.png)Source

Machine learning is the most fun if you can apply it to new types of problems without just being restricted to any set of feature extractors provided you. The process of building a good model can also become much faster and easier if you start with good features that are appropriate to whatever you’re trying to teach the learning algorithm.

Generating useful inputs for machine learning will focus on features, how to build, select and process features in order to get good results from your learning algorithms. As we’ve seen previously, a feature is a measurement of some phenomenon represented as a number.

[Source](https://cdn.hashnode.com/res/hashnode/image/upload/v1632827286659/tKKKoCodl.html)Source

In machine learning we typically create models where each data point is represented using more than one feature. That is, we have a list of several features for each training example and the same for each example our trained model sees when it’s running.

[Source](https://cdn.hashnode.com/res/hashnode/image/upload/v1632827294179/LlQ3slk3j.html)Source

This list of features is called a feature vector. We can assume the following about our feature vector. First, this feature vector or feature list is always the same length. Second, each feature in this vector can always be interpreted as meaning the same thing.

[Source](https://www.kadenze.com/courses/machine-learning-for-musicians-and-artists-v/sessions/sensors-and-features-generating-useful-inputs-for-machine-learning)Source

A typical example, if we start out collecting training examples where our first feature is hand X, second feature is hand Y, and third feature is hand Z then we’ll keep these feature types in this order forever for all other training examples and all other examples the model sees in the future.

Awesome. You made it to the end of day 028. I hope you found this informative. Thank you for taking time out of your schedule and allowing me to be your guide on this journey.

Reference

*https://www.kadenze.com/courses/machine-learning-for-musicians-and-artists-v/sessions/sensors-and-features-generating-useful-inputs-for-machine-learning*