Video Lectures Help
Having trouble viewing lectures? Try changing players. Your current player format is html5. Change to flash.
- Why do we need machine learning? [13 min]
- What are neural networks? [8 min]
- Some simple models of neurons [8 min]
- A simple example of learning [6 min]
- Three types of learning [8 min]
-
Types of neural network architectures [7 min]
Slides for Types of neural network architectures [7 min]Slides (pdf) for Types of neural network architectures [7 min]Subtitles (text) for Types of neural network architectures [7 min]Subtitles (srt) for Types of neural network architectures [7 min]Video (MP4) for Types of neural network architectures [7 min]
- Perceptrons: The first generation of neural networks [8 min]
- A geometrical view of perceptrons [6 min]
- Why the learning works [5 min]
- What perceptrons can't do [15 min]
-
Learning the weights of a linear neuron [12 min]
Slides for Learning the weights of a linear neuron [12 min]Slides (pdf) for Learning the weights of a linear neuron [12 min]Subtitles (text) for Learning the weights of a linear neuron [12 min]Subtitles (srt) for Learning the weights of a linear neuron [12 min]Video (MP4) for Learning the weights of a linear neuron [12 min]
- The error surface for a linear neuron [5 min]
- Learning the weights of a logistic output neuron [4 min]
- The backpropagation algorithm [12 min]
- Using the derivatives computed by backpropagation [10 min]
- Learning to predict the next word [13 min]
- A brief diversion into cognitive science [4 min]
- Another diversion: The softmax output function [7 min]
- Neuro-probabilistic language models [8 min]
-
Ways to deal with the large number of possible outputs [15 min]
Word Map for Ways to deal with the large number of possible outputs [15 min]Subtitles (text) for Ways to deal with the large number of possible outputs [15 min]Subtitles (srt) for Ways to deal with the large number of possible outputs [15 min]Video (MP4) for Ways to deal with the large number of possible outputs [15 min]
-
Why object recognition is difficult [5 min]
Lecture 5 slides in pptx for Why object recognition is difficult [5 min]Lecture 5 slides in pdf for Why object recognition is difficult [5 min]Subtitles (text) for Why object recognition is difficult [5 min]Subtitles (srt) for Why object recognition is difficult [5 min]Video (MP4) for Why object recognition is difficult [5 min]
- Achieving viewpoint invariance [6 min]
- Convolutional nets for digit recognition [16 min]
-
Convolutional nets for object recognition [17min]
(hard) Gradient-based learning applied to document recognition for Convolutional nets for object recognition [17min]Convolutional networks for images, speech, and time series for Convolutional nets for object recognition [17min]Subtitles (text) for Convolutional nets for object recognition [17min]Subtitles (srt) for Convolutional nets for object recognition [17min]Video (MP4) for Convolutional nets for object recognition [17min]
-
Overview of mini-batch gradient descent
Lecture 6 slides in pptx for Overview of mini-batch gradient descentLecture 6 slides in pdf for Overview of mini-batch gradient descentSubtitles (text) for Overview of mini-batch gradient descentSubtitles (srt) for Overview of mini-batch gradient descentVideo (MP4) for Overview of mini-batch gradient descent
- A bag of tricks for mini-batch gradient descent
- The momentum method
- Adaptive learning rates for each connection
- Rmsprop: Divide the gradient by a running average of its recent magnitude
- Modeling sequences: A brief overview
- Training RNNs with back propagation
- A toy example of training an RNN
- Why it is difficult to train an RNN
- Long-term Short-term-memory
-
A brief overview of Hessian Free optimization
Lecture 8 slides in pptx for A brief overview of Hessian Free optimizationLecture 8 slides in pdf for A brief overview of Hessian Free optimizationSubtitles (text) for A brief overview of Hessian Free optimizationSubtitles (srt) for A brief overview of Hessian Free optimizationVideo (MP4) for A brief overview of Hessian Free optimization
- Modeling character strings with multiplicative connections [14 mins]
-
Learning to predict the next character using HF [12 mins]
Generating Text with Recurrent Neural Networks for Learning to predict the next character using HF [12 mins]Subtitles (text) for Learning to predict the next character using HF [12 mins]Subtitles (srt) for Learning to predict the next character using HF [12 mins]Video (MP4) for Learning to predict the next character using HF [12 mins]
- Echo State Networks [9 min]
-
Overview of ways to improve generalization [12 min]
Lecture 9 slides in pptx for Overview of ways to improve generalization [12 min]Lecture 9 slides in pdf for Overview of ways to improve generalization [12 min]Subtitles (text) for Overview of ways to improve generalization [12 min]Subtitles (srt) for Overview of ways to improve generalization [12 min]Video (MP4) for Overview of ways to improve generalization [12 min]
- Limiting the size of the weights [6 min]
- Using noise as a regularizer [7 min]
- Introduction to the full Bayesian approach [12 min]
- The Bayesian interpretation of weight decay [11 min]
- MacKay's quick and dirty method of setting weight costs [4 min]
-
Why it helps to combine models [13 min]
lecture 10 slides in pptx for Why it helps to combine models [13 min]lecture 10 slides in pdf for Why it helps to combine models [13 min]Subtitles (text) for Why it helps to combine models [13 min]Subtitles (srt) for Why it helps to combine models [13 min]Video (MP4) for Why it helps to combine models [13 min]
- Mixtures of Experts [13 min]
- The idea of full Bayesian learning [7 min]
- Making full Bayesian learning practical [7 min]
- Dropout [9 min]
- Hopfield Nets [13 min]
- Dealing with spurious minima [11 min]
- Hopfield nets with hidden units [10 min]
- Using stochastic units to improv search [11 min]
- How a Boltzmann machine models data [12 min]
- Boltzmann machine learning [12 min]
- OPTIONAL VIDEO: More efficient ways to get the statistics [15 mins]
- Restricted Boltzmann Machines [11 min]
- An example of RBM learning [7 mins]
- RBMs for collaborative filtering [8 mins]
-
The ups and downs of back propagation [10 min]
lecture 13 slides in pptx for The ups and downs of back propagation [10 min]Lecture 13 slides in pdf for The ups and downs of back propagation [10 min]Subtitles (text) for The ups and downs of back propagation [10 min]Subtitles (srt) for The ups and downs of back propagation [10 min]Video (MP4) for The ups and downs of back propagation [10 min]
- Belief Nets [13 min]
- Learning sigmoid belief nets [12 min]
- The wake-sleep algorithm [13 min]
-
Learning layers of features by stacking RBMs [17 min]
Self-taught learning: transfer learning from unlabeled data for Learning layers of features by stacking RBMs [17 min](easy) To recognize shapes, first learn to generate images for Learning layers of features by stacking RBMs [17 min](hard) A fast learning algorithm for deep belief nets for Learning layers of features by stacking RBMs [17 min]lecture 14 slides in pptx for Learning layers of features by stacking RBMs [17 min]Lecture 14 slides in pdf for Learning layers of features by stacking RBMs [17 min]Subtitles (text) for Learning layers of features by stacking RBMs [17 min]Subtitles (srt) for Learning layers of features by stacking RBMs [17 min]Video (MP4) for Learning layers of features by stacking RBMs [17 min]
- Discriminative learning for DBNs [9 mins]
- What happens during discriminative fine-tuning? [8 mins]
- Modeling real-valued data with an RBM [10 mins]
- OPTIONAL VIDEO: RBMs are infinite sigmoid belief nets [17 mins]
- From PCA to autoencoders [5 mins]
- Deep auto encoders [4 mins]
- Deep auto encoders for document retrieval [8 mins]
- Semantic Hashing [9 mins]
-
Learning binary codes for image retrieval [9 mins]
Using Very Deep Autoencoders for Content-Based Image Retrieval for Learning binary codes for image retrieval [9 mins]Subtitles (text) for Learning binary codes for image retrieval [9 mins]Subtitles (srt) for Learning binary codes for image retrieval [9 mins]Video (MP4) for Learning binary codes for image retrieval [9 mins]
- Shallow autoencoders for pre-training [7 mins]
-
OPTIONAL: Learning a joint model of images and captions [10 min]
lecture 16 slides in pptx for OPTIONAL: Learning a joint model of images and captions [10 min]lecture 16 slides in pdf for OPTIONAL: Learning a joint model of images and captions [10 min]Subtitles (text) for OPTIONAL: Learning a joint model of images and captions [10 min]Subtitles (srt) for OPTIONAL: Learning a joint model of images and captions [10 min]Video (MP4) for OPTIONAL: Learning a joint model of images and captions [10 min]
- OPTIONAL: Hierarchical Coordinate Frames [10 mins]
- OPTIONAL: Bayesian optimization of hyper-parameters [13 min]
-
OPTIONAL: The fog of progress [3 min]
lecture 16 slides in pptx for OPTIONAL: The fog of progress [3 min]Lecture 16 slides in pdf for OPTIONAL: The fog of progress [3 min]Subtitles (text) for OPTIONAL: The fog of progress [3 min]Subtitles (srt) for OPTIONAL: The fog of progress [3 min]Video (MP4) for OPTIONAL: The fog of progress [3 min]