lherranz | Luis Herranz

Mix and match networks

We recently explored how we can take multiple seen image-to-image translators and reuse them to infer other unseen translations, in an approach we call

Rotating networks to prevent catastrophic forgetting

August 21, 2018February 17, 2021 continual learning, deep learning, transfer learning

In contrast to humans, neural networks tend to quickly forget previous tasks when trained on a new one (without revisiting data from previous tasks).

Generative adversarial networks and image-to-image translation

August 7, 2018August 7, 2018 deep learning, generative adversarial networks

Yet another post about generative adversarial networks (GANs), pix2pix and CycleGAN. You can already find lots of webs with great introductions to GANs (such as here,

Deep network compression and adaptation

July 28, 2018September 5, 2018 deep learning, model compression

Network compression and adaptation are generally considered as two independent problems, with compressed networks evaluated on the same dataset, and domain-adapted or task-adapted networks