Mar 2021 - Apr 2021
Required to accomplish a multi-class classification task on the provided dataset. Only use scipy, numpy and pandas
The given datasets have already been split into the training set and test set with 50000 instances and 10000 instances respectively. Each instance consists of 128 numerical attributes. The given labels in the datasets are 0 to 9, ten classes. Each class have 5000 instances for training.
The expected outcome of this project is to successfully construct an efficient multi-layer neural network model that can correctly classify instances with high accuracy.
- ReLU activation
- Weight Decay
- Momentum in SGD
- Dropout
- Softmax and Cross-Entropy Loss
- Mini-batch training
- Batch Normalisation
Datasets can be downloaded from here: https://drive.google.com/file/d/1vEev9c_PYgzjryY2Sd4gDV4CSq_DDb11/view?usp=sharing