 T 8/27: Overview; Spectrograms. HTML, Jupyter source.
 R 8/29: Fourier Transforms: MOV, WMV, MP4, Sample Problems, Solutions.
 T 9/3: Filtering. MOV, WMV, MP4, Sample Problems, Solutions.
 R 9/5: ZeroMean White Gaussian Noise. MOV, WMV, MP4, Sample Problems, Solutions.
 T 9/10: Principal Component Analysis. Slides: PPTX, PDF. Sample Problems, Solutions.
 R 9/12: Discrete Cosine Transform. Slides: PPTX, PDF.
 T 9/17: MP2 walkthrough, and introduction to robotics.
 R 9/19: Exam 1 Review
 T 9/24: Exam 1 (in class)
 R 9/26: Image filtering and Image features. PPTX, PDF. Optional supplementary material: derivation of the matched filter.
 T 10/1: Integral Image filtering, and Adaboost. Video; powerpoint; pdf. Sample Problem about integral images, and its Solution. And here is the article that proposed the name "integral image" for this type of feature: Robust RealTime Object Detection by Viola and Jones.
 R 10/3: Image upsampling, downsampling, and interpolation. IPYNB, HTML. Sample Problems, Solution.
 T 10/8: Speech Signals. MOV, WMV, MP4, Sample Problems, Solutions. (Optional extra material on windowed speech: MOV, MP4)
 R 10/10: The LPC10 Speech Coder; LPC Synthesis. powerpoint; pdf.

T 10/15: International Phonetic Alphabet. Spectrogram Reading
 Manner Class: HTML, Jupyter source
 Vowels: HTML, Jupyter source
 Semivowels: HTML, Jupyter source
 Fricatives: HTML, Jupyter Source
 Nasals: HTML, Jupyter source
 Stops and Affricates: HTML, Jupyter source
 R 10/17: Exam 2 Review (Practice Exam and its Solutions.)
 T 10/22: Exam 2

R 10/24: Hidden Markov Models. I'm going to deliver this
lecture, this year, using the blackboard. The Rabiner
tutorial article is the most useful reference.
 Lawrence Rabiner, A Tutorial on Hidden Markov Models and Selected Applications in Speech Recognition, Proceedings of the IEEE, 1989.
 A tutorial on Bayesian classification using Gaussian pdfs: PPTX, PDF. Sample Problems, Solutions.

An ipython notebook using HMMs to recognize which city
you're in, based on monthly temperature readings for
the last year:
IPYNB,
HTML.
 An ipython notebook that generates training data from an HMM, and then trains a second HMM using the training data. The goal is for the second HMM to match the first HMM, as closely as possible. IPYNB, HTML.
 Speech recognition details: (1) mapping tokens to types, (2) Gaussian surprisal, (3) scaled forwardbackward algorithm, (4) Viterbi algorithm.
 R 10/31: MP5 walkthrough. PPTX, PDF.

T 11/5: Face animation by moving pixels.
 In class I'll use the slides by Vuong Le. If you'd like another view on the same data, here are slides I wrote about affine transforms and barycentric coordinates.
 Sample Problems, Solutions.

R 11/7: Introduction to Artificial Neural Nets.
 In class I'll use the slides by Vuong Le. If you'd like another view on the same data, here are slides I wrote about neural nets.
 Sample Problems, Solutions.
 T 11/12: MP6 walkthrough
 R 11/14: Face Animation: State of the art/current methods (Kuangxiao Gu)
 T 11/19: Recurrent Neural Nets. Slides, Sample Problems, Solutions.
 R 11/21: LSTM.
 T 11/26: Vacation
 R 11/28: Vacation
 T 12/3: Convolutional Neural Nets. Slides, Sample Problems, Solutions.

R 12/5: Simulated Annealing, MiniBatch, Data Augmentation.
Slides,
a
video about simulated annealing.
There is no new theory, in this lecture, that you need for the exam. You don't need to know about simulated annealing, or about minibatch, or about data augmentation. However, the sample problem only covers (1) forwardprop, (2) knowledgebased design, and (3) backprop training, so the sample problem is fair game for the exam. Here it is: Sample Problem, Solutions.  T 12/10: Exam 3 Review
Lectures subject to change, up until the day of the lecture.