Loss functions and metrics to monitor the accuracy when training a model in Deep Learning

1. Types of loss functions
The loss is the penalty for failing to achieve a desired value.
There are 2 major types:
 - Regression losses
 - Classification losses
2. Loss functions
For Regression:
 - Mean Squared Error Loss
 - Mean Squared Logarithmic Error Loss
 -  Mean Absolute Error Loss
For Binary Classification:
 - Binary Cross-Entropy
 - Hinge Loss
 - Squared Hinge Loss
For Multi-Class Classification:
 - Multi-Class Cross-Entropy Loss
 - Sparse Multiclass Cross-Entropy Loss
 - Kullback Leibler Divergence Loss
3. Keras metrics to monitor the accuracy when training a model
For classification problems:
 - Binary Accuracy: binary_accuracy, acc
 - Categorical Accuracy: categorical_accuracy, acc
 - Sparse Categorical Accuracy: sparse_categorical_accuracy
 - Top k Categorical Accuracy: top_k_categorical_accuracy
 - Sparse Top k Categorical Accuracy: sparse_top_k_categorical_accuracy
 - Custom Metrics
For regression problems:
 - Mean Squared Error: mean_squared_error, MSE or mse
 - Mean Absolute Error: mean_absolute_error, MAE, mae
 - Mean Absolute Percentage Error: mean_absolute_percentage_error, mape
 - Cosine Proximity: cosine_proximity, cosine
 - Custom Metrics