Update of 2020: we hired several excellent researchers, but still have open positions for PhD students and postdocs on the European Research Council (ERC) Advanced Grant on "Recurrent Neural Networks and Related Machines That Learn Algorithms" (AlgoRNN, compare our NIPS Symposium on this topic).

AlgoRNN blurb: Recurrent neural networks (RNNs) are general parallel-sequential computers. Some learn their programs or weights. Our LSTM was the first RNN to win machine learning contests, and is now used billions of times per day through the world's most valuable public companies, e.g., for speech recognition on over 2 billion smartphones, language translation, etc. Nevertheless, in lots of real-world tasks RNNs do not yet live up to their full potential. Although universal in theory, in practice they fail to learn certain important types of algorithms. This ERC project will go beyond today's best RNNs through novel RNN-like systems that address some of the biggest open RNN problems and hottest RNN research topics including attention & fast weights, RNN-based reinforcement learning, optimal transfer learning, and metalearning.

We have direct access to one of the fastest supercomputers in the world, with over 5000 GPUs, also based in Lugano, Switzerland.

We are looking for outstanding candidates with experience in topics such as deep learning neural networks, video analysis, RNNs and Long Short-Term Memory (LSTM), meta-learning or "learning to learn learning algorithms", evolutionary computation, RNN evolution, compressed network search, reinforcement learning (RL) for robots in realistic partially observable environments where traditional RL (for board games etc) does not work, unsupervised learning through adversarial neural networks that fight each other in a minimax game, curiosity-driven learning and the Formal Theory of Fun & Creativity, hierarchical reinforcement learning, general artificial intelligence, universal learning machines, search for programs running on general purpose computers such as RNNs, and "learning to think." (You will also collaborate with other lab members on other projects - we are one big family!)

The general goal is to advance the state of the art in machine learning and AI. Duration for postdocs: initially for 2 years, with possibility of prolongation.

We also have additional job openings with separate instructions thanks to generous gifts by Google DeepMind and NVIDIA (which also gave us a DGX-1 deep learning supercomputer), and an SNF grant on deep learning-based video analysis (we have filled positions for the RNNAIssance project).

We don't want to rush it, and will take our time to find excellent collaborators.

PhD student salary: Initially CHF 46'000 per year. Standard postdoc salary: CHF 80'000 per year. Low Swiss taxes. There is travel funding in case of papers accepted at important conferences. There are also exciting connections to the world of AI startups.

There is an opportunity to participate in teaching courses (in English) on machine learning, in Switzerland's first A.I. Master's degree program at the University of Lugano. Official language at IDSIA is English.

INSTRUCTIONS: Submit your CV, a brief statement of research interests explaining why exactly you'd like to work with us, and a list of 3 references and their email addresses to rnnai@idsia.ch. (Do not send articles or transcripts or other large files - they will be deleted.) In the subject header, mention your full name followed by the keyword erc, and add "PhD" in case of PhD students, or add "postdoc" in case of postdocs. For example, if your name is Jo Mo, and you are applying as a PhD student, use subject: Jo Mo erc PhD. We will continually evaluate applications until all positions are filled, which may take a while - we will occasionally update this web site.