VS298 (Fall 06): Suggested projects

This is not an exhaustive listing, just some suggestions to get you started thinking.

• NETtalk. Train a multi-layer perceptron to convert text to speech. You can get Sejnowski & Rosenberg's original paper and the data they used here. (You will need a DECtalk speech synthesizer to play the phonemes - you can probably pick up a used one online.)

• Recognition of handwritten digits. Train a MLP to classify handwritten digits 0-9. You can get some training data here. You may wish to follow the convolutional network methodology of Yann LeCun (try the simpler, earlier model), or invent your own method.

• Sparse coding and decorrelation. Implement Peter Foldiak's network and train it on the handwritten digits above to learn the features of this data. You may wish to then try supervised learning on the learned features to see if it has simplified the classification problem.

• Cortical maps. The elastic net model of Durbin and Mitchison is typical of many cortical map models in that they learn directly on a parameterized feature space. But the cortex simply gets a bunch of inputs from the LGN, and so it needs to learn features such as orientation at the same time as it organizes them into a feature map. How would you go about learning a feature map for orientation position directly from simulated LGN inputs? (You may wish to consult the book of Risto Miikkulainen for recent efforts in this area.)

• Feedforward vs. recurrent weights. As we discussed in class, one can implement a given input-output mapping in a neural network using just feedforward weights: <math> y = W x</math>, or using just recurrent weights: <math> \tau dy/dt + y = x + M y</math>, or both: <math>\tau dy/dt + y = W x + M y</math>. Probably there is a trade-off here in terms of minimizing overall wiring length and settling time - i.e., feedforward networks are fast but require lots of synapses, while recurrent networks are slower but can implement more complex functions with local connections. Explore these tradeoffs for a particular problem - e.g., implementing an array of Gabor filters in model of V1.

• Sparse codes and associative memory. Associative memory models have the highest storage capacity for sparse patterns (i.e., patterns with many zeros). At the same time, recent work has shown how it is possible to convert natural images and sounds into a sparse format, and there is some evidence for this happening in the brain.

• Hierarchical restricted Boltzmann machines.

• Integrate-and-fire model neurons.