Image classification from 10 classes (fish, dog, device, chainsaw, church, horn, truck, petrol, golf, parachute). Very similar to CIFAR-10 but different set of classes.
- Start with a randomized neural network (not allowed to use pre-trained model)
- Time limit to train is 120 seconds, excluding data preprocessing and model evaluation
- Test set cannot be used during training
Recommendation system for animes given user preferences. Similar problems available on Kaggle.
- Start with a non-factor model approach
- Check if factor model approaches or other methods discuss in class can help improve the system
- Optionally improve the recommendation system by all means possible
The data can be obtained from Kaggle.
Given cities.npy as a TSP instance, find the optimum tour.
None.