Coursera Deeplearning.ai : Improving Neural Networks Certification

Second achievement from the Coursera Deeplearning.ai Deep Learning specialization: I just completed the course on "Improving Neural Networks", on which I acquired and solidified foundations on concepts such that:

1. Regularization

Bias and variance deep neural networks, L1 and L2 regularization on logistic regression and deep l-layer feed-forward networks, dropout regularization, early stopping, input normalization, vanishing and exploding gradients, smarter weight initializations of deep networks' parameters, gradient checking. 

2. Optimization 

Batch, mini-batch and stochastic gradient descent, exponentially weighted averages, momentum, RMSprop, Adam optimizer, local optima in neural networks cost functions. 

3. Hyperparameter Tuning  

Grid search, randomized sampled search, "panda" and "caviar" training school of thoughts (c.f. Andrew Ng), batch normalization. 

4. Multi-class classification and deep learning frameworks

Softmax regression, Tensorflow "from principles" deep neural network building for multi-class classification. 

Certificate: https://www.coursera.org/account/accomplishments/verify/77R793YC7Y4J?utm_source=link&utm_campaign=copybutton_certificate&utm_product=course 

#machinelearning #datascience #deepnetoworks #neuralnetworks #optimization #regularization #hyperparametertuning #learning #ai