Research internship (summer 2020)
Potential internship topics include:
- Scalable uncertainty estimation techniques in neural networks towards addressing model over-confidence;
- Advance methods for combining meta-learning/few-shot learning and continual learning;
- Representation learning for downstream tasks, e.g. improving data efficiency, controllable generation, model interpretability, robustness, etc.
Interviewing process at MSRC include both ML research questions and ML engineering questions.
If you are interested, send me an email at Firstname.Lastname [at] company [dot] com. Please include your CV and a short paragraph stating your research experience.
I am very happy to collaborate with academia people (as long as it doesn't touch company data, and it doesn't result in IP conflicts).
I have worked on, and continue to work on, the following topics:
- Approximate inference, e.g. variational inference, message passing, SG-MCMC, function-space inference;
- Applications of uncertainty quantification, e.g. continual learning, adversarial attack & defence;
- Generative modelling, e.g. VAE/GAN/flow-based models/energy-based models/implicit models;
- Others, e.g. meta-learning, transfer learning, continual/life-long learning, disentangled representations, gradient estimation.
I am open to collaborate with people that have very different background. E.g. one of my on-going collaboration is on Bayesian methods applied to NLP tasks (and I'm still learning there).
If you are interested, send me an email at liyzhen2 [at] gmail [dot] com.