The Neural and Machine Learning group at the University of Bristol, UK is looking for a PhD student to work at the intersection between computational neuroscience and machine learning (funded by a EPSRC Doctoral Training grant). We are part of the Bristol Computational Neuroscience Unit and are associated with the world-class neuroscience and machine learning communities at Bristol, and we also collaborate with other national and international labs. Applicants should hold or expect to obtain a First Class or Upper Second Class Honours degree or equivalent in engineering, computer science, physics, mathematics, computational neuroscience or machine learning (or equivalent), and a strong interest in neural and biological systems. This position is open to applicants who are UK citizens, or EU citizens resident in the UK for at least 3 years.

Project: Deep temporal credit assignment with dynamic synapses

Successful behaviour relies on the ability of animals and machines to predict the outcome of their actions. Selecting which synapses (or parameters) to adjust to trigger the right action is a hard problem as it depends on past sensory information. Exactly how does the brain solve such temporal credit assignment problems has remained largely enigmatic. In this project, the student will developed a novel theoretical framework to address the temporal credit assignment problem. This framework will be compared with experimental observations and tested in machine learning tasks in collaboration with experimental neuroscientists and machine learning researchers.

Relevant papers:

• Llera M, Sacramento J and Costa RP - Computational roles of plastic probabilistic synapses - Current Opinion in Neurobiology, Volume 54, Feb 2019
• Sacramento J, Costa RP, Bengio Y and Senn W - Dendritic cortical microcircuits approximate the backpropagation algorithm - 32nd Conference on Neural Information Processing Systems (NeurIPS 2018), Montreal, Canada.
• Costa RP, Assael YM, Shillingford B, Freitas N and Vogels TP - Cortical microcircuits as gated-recurrent neural networks - 31st Conference on Neural Information Processing Systems (NIPS 2017), Long Beach, CA, USA.
• Costa RP, Froemke RC, Sjöström PJ and van Rossum MCW - Unified pre- and postsynaptic long-term plasticity enables reliable and flexible learning - eLife 2015; 10.7554/eLife.09457

University of Bristol:

The University of Bristol is a world-class university ranked among the top 10 universities in the UK, and top 50 in the world. The university has one of the strongest neuroscience and machine learning communities in the UK (e.g.: Visual Information Lab, Intelligent Systems Lab and Institute for Statistical Science). The university also runs the unique Wellcome Trust Doctoral Program in Neural Dynamics with PhD students working at the intersection between theoretical and experimental neuroscience, and is associated with the Alan Turing Institute for data science and artificial intelligence. The city itself has been ranked as the best place to live in the UK and one of the most cool destinations in the world according to National Geographic.

Details on how to apply:

http://www.bristol.ac.uk/engineering/postgraduate/funding/epsrcdta.html

When applying online please follow the instructions in the link above and include a CV, two references and a motivation letter.

Deadline: 28^th of February 2019

Best wishes,

Rui Ponte Costa

Lecturer (Assistant professor) in Computational Neuroscience and Machine Learning

University of Bristol, UK

Visiting Research Scientist at University of Oxford/Bern

http://ruipcosta.weebly.com