Karen Simonyan
Karen Simonyan co-authored the VGG research with Andrew Zisserman in 2014, studying deep convolutional networks built with small filters. In 2016 he was a co-author of WaveNet, an autoregressive model that generates raw audio samples. These papers contributed to visual and audio learning systems. Microsoft’s 19 March 2024 announcement identified him as Inflection’s co-founder and chief scientist and stated that he would become chief scientist of Microsoft AI.
Timeline
-
2014
Deep networks for recognition
Simonyan and Andrew Zisserman studied the effect of network depth in large-scale image recognition. Their VGG paper used small 3×3 convolution filters and evaluated networks with 16–19 weight layers.
-
2014
ImageNet and shared representations
The VGG paper reported first place in localisation and second place in classification for the team’s 2014 ImageNet Challenge submission. It also described releasing two models for further research.
-
2016
Raw audio generation
Simonyan was one of the authors of WaveNet, first submitted in September 2016. The network predicts each audio sample from previous samples. The paper reported text-to-speech and other audio experiments. This added generative audio modelling to his visual research.
-
2024
An announced Microsoft AI role
Microsoft’s 19 March 2024 announcement stated that Simonyan would join Microsoft AI as chief scientist. It identified him as Inflection’s co-founder and chief scientist.
Sources and credits
Photo: chessprogramming.org; Karen Simonyan - Chess Programming Wiki; creator not stated. License: Not stated by source. Source: chessprogramming.org. Cropped to a square and resized. Ask us to remove or replace this photo.