Machine Learning on a Cancer Dataset - Part 23
In the fifth tutorial on neural networks with scikit-learn we look at how our multi-layer perceptron learns.
Similar to what we've done for decision trees and random forests, we're going to look at the importance each feature carries in the decision making process. But this is not as easy because we're trying to visualize the weight matrix over 100 hidden units and 30 input features. And that's a bit complex.
We're using matplotlib and an insightful colormap to visualize and try to make sense of the whole 'thing'. Inferring interpretation and driving insight from it should be done carefully. Please see the video tutorial for the full scoop.
As a reminder:
In this series I'm going to explore the cancer dataset that comes pre-loaded with scikit-learn. The purpose is to train the classifiers on this dataset, which consists of labeled data: ~569 tumor samples, each labeled malignant or benign, and then use them on new, unlabeled data.
Previous videos in this series:
- Machine Learning on a Cancer Dataset - Part 20
- Machine Learning on a Cancer Dataset - Part 21
- Machine Learning on a Cancer Dataset - Part 22
This is very interesting, but I need to start back from Part 1...
Very informative. Thank you!
This makes me optimistic for the future.
Thanks for sharing,
followed.
thanks!
Very interesting. Have you looked into using Watson Machine Learning from IBM Data Science Experience for future machine learning tutorials?
yes, I did a few tutorials for watson in the past, but not specific to ML. will probably get to that in the future. thanks for the reminder.
Awesome, looking forward to it! I think cognitive machine learning apps are the way to go
AI is the not the onto-logically right word - machine learning even. it's all about patterns and data processing - as you consider this you can do any application you want