This repository contains my assignments from the Neural Networks - Deep Learning university course (NDM-07-05). The course covered topics such as neural networks, support vector machines, and radial basis function neural networks.
- The MNIST digit classification using MLP.
Experiments related to the learning part, as it was described and analyzed in the lectures of the course:
- Comparison with KNN, Nearest Class Centroid.
- Weight initialization.
- Standardization-Normalization.
- One Hot Encoding.
- Batch, Mini-Batch, Stochastic Gradient Descent.
- Data Shuffling.
- Cross Validation for hyper-parameter tuning.
- Committees of Neural Networks.
- Recognition of odd and even numbers in the decimal digits (0,1,…,9) of MNIST.
Experiments related to the learning part, as it was described and analyzed in the lectures of the course:
- Comparison with KNN, Nearest Class Centroid.
- Standardization-Normalization.
- Regularization Parameter(C) & Kernels(eg Polynomial).
- Consecutive Cross Validations for hyper-parameter tuning.
- Training with the best set of parameters.
- Primal Vs Dual Space.
- The MNIST digit classification using RBF NNs.
Experiments related to the learning part, as it was described and analyzed in the lectures of the course:
- Comparison with KNN, Nearest Class Centroid and MLPs.
- Different Standard Deviations for the Gaussian function.
- Different Centroid Initializations for the Gaussian function.
- Extreme Learning Machine (ELM).