L09.3.5: A Mini Xception-like Model Example

This video is really a conclusion to the previous one where Iintroduced a few modern best practices and concepts for creating convolutional neural networks. In particular I talked about residual connections, batch normalization and depthwise separable convolutions. This video shows an example of putting those elements along with all of the best-practices for convolutional network architectures that we discussed into practice. I walk you through creating an Xception-like model, that uses the same architecture design and new elements we discussed. I show that this new model when trained by scratch again on the cats v dogs dataset will show a noticeable improvement in performance over a convolutional model that does not use the new practices we introduced. Resources: Textbook: Chollet (2022). "Deep Learning with Python (2ed)". Manning. https://www.amazon.com/dp/1617296864/?bestFormat=true&k=deep%20learning%20with%20python&ref_=nb_sb_ss_w_scx-ent-pd-bk-d_de_k0_1_15 CSci 560 Class Repository: https://github.com/csci560-nndl/nndl Contains video slides and iPython notebooks for this course. 00:00 Introduction 01:33 Mini Xception-like model architecture 08:26 Train and evaluation Xception-like model on cat v dog 11:25 Summary