WEEK 6— MUSIC GENRE CLASSIFICATION
Hello, this is our sixth blog post for our Fundamentals of Machine Learning course project. This week, we increased our input samples, used pca for feature selection, and added one more hidden layer to our neural network model. Let’s begin talking about these changes.
We tried different methods in our network but we saw that we cannot improve accuracy our model cannot learn enough. We decided to develop our dataset by converting 30-second songs into 10 3-second songs. And it was the right step for our aim. Our accuracy directly increased. After that, we decided to develop our model, we had 32 different features for classification. We decided to use pca, it effects models to become more efficient as the reduced feature set boosts learning rates and diminishes computation costs by removing redundant features. We used pca and we had 11 features which are affected mostly the label. Then, as we mentioned in the previous blog, we started training with a neural network model. There added one more layer in our network, so we have 5 layers in our network. The first one is an input layer. Then there are 3 hidden layers and the last layer is an output layer. The output layer has 10 neurons as we are classifying it into 10 genres.
We used Relu as an activation function and used adam algorithm which is largely used in deep learning. Our model’s goal is to reduce sparse categorical cross-entropy loss function. We used the fit function with 30 epochs and 128 batch sizes. And with these properties, accuracy with these different features is equal to 91%.
NEXT WEEK
We tried and got good results with a deep neural network, there are more different classification methods. Next week we will try different methods, share their results, make the last touches, and finalize our project.
WRITERS’ MEDIUM PAGES
