Staff research scientist at InstaDeep
A researcher who developed connectionist temporal classification for LSTMs and created neural Turing machines.
Alex Graves is a computer scientist who earned a BSc in theoretical physics from the University of Edinburgh and a PhD in artificial intelligence from the Technical University of Munich, supervised by Jürgen Schmidhuber at the Dalle Molle Institute for Artificial Intelligence Research. He then worked as a postdoc with Schmidhuber in Munich and with Geoffrey Hinton at the University of Toronto.
At the Dalle Molle Institute he trained long short-term memory networks with a method called connectionist temporal classification (CTC), which outperformed traditional speech recognition models in some applications. In 2009 his CTC-trained LSTM became the first recurrent neural network to win pattern recognition contests, taking several handwriting recognition competitions. Google uses CTC-trained LSTM for speech recognition on smartphones. He also created neural Turing machines and the differentiable neural computer, and wrote the 2023 paper Bayesian Flow Networks. He is a staff research scientist at InstaDeep and previously worked at Google DeepMind in London.
Trained LSTM networks with CTC, a method that outperformed traditional speech recognition models in some applications and is used by Google for smartphone speech recognition.
In 2009 his CTC-trained LSTM became the first recurrent neural network to win pattern recognition contests, winning several connected handwriting recognition competitions.
Created neural Turing machines and the closely related differentiable neural computer, networks coupled to external memory.