Brief to Generative and Discriminative models Learning approaches can be broadly divided into two categories; generative and discriminative. Generative models learn the underlying distribution of data representing various categories for example, model may learn distribution of pixels in an image when a cat is present differently in comparison to presence...
[Read More]
Checkpointing and evaluating trained model in Tensorflow
Brief on why model checkpointing is even needed?
[Read More]
Understanding GANs; analyzing Generator and Discriminator losses
GANs; known as Generative Adversarial Networks proposed in 2014 by Ian Goodfellow for their ability to learn a function mapping a random distribution to desired. Given this mapping, the function can synthetically generate desired number of observations. Starting from a random distribution or a given prior the generator adjusts its...
[Read More]