This Friday we'll have two talks followed by drinks.
16.00: Nal Kalchbrenner (Google Brain Amsterdam) -- Generating Our World One Bit At a Time
16.30: Ilya Markov (U. Amsterdam) -- Visual Features for Learning to Rank
Nal Kalchbrenner (Google Brain Amsterdam) -- Generating Our World One Bit At a Time
I will discuss the very latest developments in sequential generative models, a state-of-the-art unifying framework for generating all sorts of data including images, audio and text.
Ilya Markov (U. Amsterdam) -- Visual Features for Learning to Rank
Standard learning to rank (LTR) approaches combine multiple types of features, namely content-based, link-based, and user-based. Recently, visual features have been introduced in LTR. These features are extracted from visual snapshots of entire documents or their snippets using neural networks. Hand-crafted representations of documents could potentially be replaced by visual representations learned automatically from visual appearance of information. I present survey developments in the area of visual features for LTR and discuss their limitations and potentials.