Postdoc Fellow on Automatic Machine Learning

Applications are invited for a postdoc position in data mining to join Eindhoven University of Technology. There is no application deadline for this position.

Challenging post-doctoral fellowship on automated machine learning (AutoML). We are building an AutoML playground (an ‘AutoML Gym’) to train AutoML systems on many different problems and get increasingly better over time. This work is part of an Amazon Research Award.

We are seeking a highly creative and motivated post-doctoral fellow to join the Data Mining Group at the Eindhoven University of Technology. The candidate will be working in collaboration with Dr. ir. Joaquin Vanschoren, as well as the OpenML core team and Amazon Research.

The field of automated machine learning (AutoML) aims to automatically build machine leaerning models in a data-driven, objective, and automatic way. The user simply provides data, and the AutoML system automatically determines the approach that performs best for this particular application. Although the field is moving fast and substantial progress has been made, it is also hindered by a lack of benchmarking and a lack of environments for experimentation and analysis.

We are building an AutoML playground (an ‘AutoML Gym’) to train AutoML systems on many different problems and get increasingly better over time. Similar to the OpenAI Gym, which trains reinforcement learning agents on many different scenario’s, the AutoML Gym will train and test many different AutoML systems (agents) on many challenging problems. We will continuously track the performance of the AutoML agents, and store this information in a meta-data repository, a shared memory that can be accessed by any AutoML agent to perform meta-learning and become increasingly better over time.

The project will progress in three phases, first covering hyperparameter optimization to automatically construct and optimize machine learning pipelines as efficiently as possible. Next, we enable meta-learning, by giving AutoML agents access to a shared memory of prior experiments. Finally, we will include AutoML techniques that design neural architectures, covering both few-shot learning and neural architecture search.

This work is funded by an Amazon Research Award. It will be set in an very interactive environment, including the Eindhoven Data Mining Group, the OpenML team, the AutoML community, and Amazon research.

For any further inquiries on the content of the position, please contact Joaquin Vanschoren (j.vanschoren[at]tue.nl).

About the employment conditions you may contact the HR advisor, Mrs. Arianne Boekemea (pzwin[at]tue.nl) If you would like to apply, please send us your application by using the ‘apply for this job’-button at the top of this page (applications via e-mail will not be accepted).

You can upload only 5 documents up to 2 MB each. If necessary please combine documents.

Deadline: As soon as possible
Apply Link:http://tinyurl.com/y42ydpcw

Tags: postdoc position in data mining, data mining postdoc position, postdoc position in netherlands, postdoc position at eindhoven university of technology

Copyright © 2019 FindScholarship.net. All Right Reserved.