CSCI 567: Machine Learning
Copyright By PowCoder代写 加微信 powcoder
Lecture 7, Oct 20
Administrivia
• Quiz 1 grades will be released soon.
• Linear algebra tip: Whenever you see or write a matrix-matrix or matrix-
vector product, double check to make sure the dimensions match.
Make sure none of your linear-algebra operations are caught by the “matrix police”…
Administrivia
• Quiz 1 grades will be released soon.
• Linear algebra tip: Whenever you see or write a matrix-matrix or matrix-
vector product, double check to make sure the dimensions match.
• Project details will be released in 1-2 weeks (Kaggle competition).
• Groups of 4 (start forming groups)
• Today’s plan:
• Convolutional neural networks
• Sequential prediction, Markov models and (a bit of) recurrent neural
Convolutional
Neural Networks
Not much math in this part, but there’ll be empirical intuition (and cat pictures J)
The materials in this part borrow heavily from the following sources:
• Stanford’s CS231n: http://cs231n.stanford.edu/
• Deep learning book by Goodfellow, Bengio and Courville: http://deeplearningbook.org
Both website provides a lot of useful resources: notes, demos, videos, etc.
Acknowledgements
http://cs231n.stanford.edu/
http://deeplearningbook.org/
)HL�)HL�/L� �-XVWLQ�-RKQVRQ� �6HUHQD�
6Qhk