We learn Deeply.
- 01: Linear Regression of Sinewave with Gaussian Basis Functions
Perform linear regression of a noisy sinewave using a set of gaussian basis functions with learned location and scale parameters. Model parameters are learned with stochastic gradient descent. Use of automatic differentiation is required. - 02: Spiral Classification with Multi-Layer Perceptron
Perform binary classification on the spirals dataset using a multi-layer perceptron. You must generate the data yourself. - 03: : MNIST Classification
Classify MNIST digits with a (optionally convoultional) neural network. Get at least 95.5% accuracy on the test test. Challenge: 80% accuracy, with fewest number of learned parameters - 04: CIFAR 10/100
Classify CIFAR10. Acheive performance similar to the state of the art. Classify CIFAR100. Achieve a top-5 accuracy of 90%. - 05: Classify the AG News dataset
Consider the AG News dataset. which contains headlines and descriptions for a large set of news articles. Perform proper cross validation.