A few demos how to use deep learning for classification of small data sets for marketing and cyber-security
Here below a list of resources to complement the presentation, about Deep Learning, Supervised Data Science, Classification, and the reference to the dataset used
Best Regards Natalino Busa
Feel free to contact me via linkedin:
https://www.linkedin.com/in/natalinobusa
FAQ's on Deep Learning:
http://www.faqs.org/faqs/ai-faq/neural-nets/part1/preamble.html
-
scaling inputs:
http://www.faqs.org/faqs/ai-faq/neural-nets/part2/section-16.html -
how much data is needed to train a neural network?
https://www.quora.com/How-much-data-is-enough-to-train-a-deep-NN-model -
small data
https://medium.com/@ShaliniAnanda1/an-open-letter-to-yann-lecun-22b244fc0a5a#.ngpal1ojx -
fast activation functions
https://www.quora.com/What-is-hard-sigmoid-in-artificial-neural-networks-Why-is-it-faster-than-standard-sigmoid-Are-there-any-disadvantages-over-the-standard-sigmoid
Deep Learning in Python: Keras vs Lasagne vs Theano
https://www.youtube.com/watch?v=E92jDCmJNek
Introduction to Keras
https://youtu.be/Tp3SaRbql4k
Random Forests vs Deep Learning
http://blog.kaggle.com/2012/11/01/deep-learning-how-i-did-it-merck-1st-place-interview/
AI deeplearning in one graph
https://youtu.be/LcfLo7YP8O4
Challenges about deep learning and machine learning
https://youtu.be/CLDisFuDnog
deep learning introduction
https://youtu.be/15h6MeikZNg
cost function and back propagation:
https://www.youtube.com/watch?v=FYgsztDxSvE
https://www.youtube.com/watch?v=-YRB0eFxeQA
... in fact worth to follow the full course :)
Statistical view of neural networks as hierarchical/recursive glm
http://blog.shakirm.com/2015/01/a-statistical-view-of-deep-learning-i-recursive-glms/
Datasets:
Churn, artificial, based on actual data: http://www.sgi.com/tech/mlc/db/churn.names
Customer churn data: The MLC++ software package contains a number of machine learning data sets. The "churn" data set was developed to predict telecom customer churn based on information about their account. The data files state that the data are "artificial based on claims similar to real world". These data are also contained in the C50 R package. http://appliedpredictivemodeling.com/data/