MENU

Amazon sets up AI science hub in Germany

Technology News |
By Nick Flaherty
AI

Amazon has teamed with Max Planck Gesellschaft (MPG) for its first Science Hub outside the US.

Amazon scientists and Max Planck researchers in Germany will work jointly on applied research in areas such as artificial intelligence and computer vision

The teams will make their findings available to the broader community via joint publications. The collaboration will include sponsored research; open research; industrial fellowships co-supervised by Max Planck and Amazon; and community events.

The hub opens doors to further scientific collaboration with Max Planck Institutes (MPI), including the MPI for Intelligent Systems, the MPI for Software Systems, the MPI for Informatics, and the MPI for Biological Cybernetics.

“We are delighted that our Cyber Valley partner, Amazon, is making a substantial increase in its commitment to basic research in Tübingen and with the Max Planck Society in general,” said Michael J. Black, former Amazon Scholar and director of the Max Planck Institute for Intelligent Systems. “With its new research facility adjacent to the Max Planck campus, Amazon has been building world-class research and development teams. The new hub builds on this to provide an opportunity for collaborative research, sponsored projects, student fellowships, and internships. The hub is a new template for cooperation that preserves the high academic standards and independence of Max Planck while exposing interested researchers to complex real-world problems at Amazon scale. We look forward to new ground breaking research coming out of this cooperation.”

Amazon initially will fund the hub with approximately €700,000 in the first year of a five-year collaboration.

“The collaboration offers young scientists insights into both basic research at the Max Planck Institutes and applied research at Amazon, for example, how computer vision technology might be leveraged to improve our Amazon Fashion customers’ shopping experiences,” said Betty Mohler Tesch, principal scientist in Amazon Fashion. “Core to the vision of the hub is the creation of open doors and the exchange of ideas through community events, open houses, and even just over coffee.”

The collaboration will also include six PhD fellowships exploring the following topics:

Robust Deep Learning: The AWS Tübingen Causal Representation Learning Team will collaborate with the department of Computer Vision and Machine Learning at the MPI for Informatics to develop new architectures and training techniques that leverage unlabeled or weakly labeled data.

Label Efficient Deep Learning: The AWS Tübingen Causal Representation Learning team will collaborate with the department of Computer Vision and Machine Learning at the MPI for Informatics to reduce the dependency on downstream labeled data for transfer learning with modern neural networks.

Scalable Feature Relevance Attribution: The AWS Tübingen Causality team will collaborate with the department of Empirical Inference at MPI for Intelligent Systems to redefine feature relevance attribution in a way that can be applied to features of different levels of abstraction.

Label Efficient Deep Learning: The AWS Tübingen Causal Representation Learning team will collaborate with the department of Autonomous Learning at the MPI for Intelligent Systems to bridge the gap between perception and reasoning in neural networks.

Merging Data Sources: The AWS Tübingen Causality team will collaborate with the Empirical Inference department at MPI for Intelligent Systems to explore how causal insights emerge from pooling statistical information that refers to different subsets of variables.

Robustness Through 3D Representations: A team from Tübingen will collaborate with MPI-INF, to investigate novel neural representations that factor into interpretable physical quantities, such as 3D shape, lighting, and appearance. The goal: make deep learning models robust to factors of variations, and to make models interpretable.

Amazon has extensive ties to MPG. Bernhard Schölkopf, Amazon distinguished scientist, is also director of the Department of Empirical Inference at the Max Planck Institute for Intelligent Systems.

Amazon also employs MPG academics as both Scholars and Visiting Academics. Current Scholars include, Bernt Schiele, Max Planck director at MPI for Informatics and professor at Saarland University who joined AWS in April 2021; and Rupak Majumdar, scientific director at the Max Planck Institute for Software Systems who joined the Automated Reasoning Group in October 2021. Michael J. Black, director at the Max Planck Institute for Intelligent Systems (MPI-IS), was an Amazon Scholar from March 2020 to December 2021. Timo Bolkart, a research scientist in the Perceiving Systems Department of the Max Planck Institute for Intelligent Systems in Tübingen, joined as a Visiting Academic in June 2021.

In late November 2019, Amazon announced it was establishing a new unit in its research center in Tübingen, Germany, a university town in central Baden-Württemberg, about 30 kilometers (19 miles) south of Stuttgart. This research team is dedicated to open research in AI, focusing on long-term challenges related to explainability, causality, and how AI systems can comprehend their environments. The Lablet is the fourth Amazon research and development centre in Germany, along with Berlin, Dresden and Aachen.

In Tübingen, Amazon is one of seven industry partners in the so-called Cyber Valley, one of Europe’s largest research collaborations between science and industry in artificial intelligence. The location enables close cooperation between industry and top researchers from all over the world. Amazon’s goals are to support this approach by investing in creating new jobs, while also promoting Tübingen as a technology location.

Amazon Fashion is seeking research proposals to advance the state of the art in clothing simulation, fit, personalization, and more. The deadline to submit proposals is June 17.

www.amazon.science

Other articles on eeNews Europe

 


Share:

Linked Articles
eeNews Europe
10s