June 2021 - Ai - Engineering Mailing Lists

FW: Postdoc fellowships available at Berkeley
by Fern, Alan Paul 17 Jun '21

17 Jun '21

This may be of interest to anyone graduating with a Ph.D. in the near future. From: Stuart RUSSELL <russell(a)berkeley.edu> Sent: Wednesday, June 16, 2021 4:07 PM To: Stuart RUSSELL <russell(a)berkeley.edu> Cc: Martin Fukui <mfukui(a)berkeley.edu> Subject: Postdoc fellowships available at Berkeley [This email originated from outside of OSU. Use caution with links and attachments.] Dear Colleague, I hope you're doing well! I am writing to let you know about the new postdoctoral Research Fellowships<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fhumancomp…> at the Center for Human-Compatible AI at UC Berkeley. If you know of any promising candidates, please either forward this email to them or let us know directly so we can reach out to them. Some more details are given below and in the official job posting<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fhumancomp…>. I'd just like to emphasize the breadth of topics we're interested in pursuing: the Center's goal is to develop provably beneficial AI systems, and that includes any and all rigorous methods for creating AI systems whose properties we understand by design. Thanks and best wishes, Stuart Russell Professor of Computer Science and Director of the Center for Human-Compatible AI<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fhumancomp…>, UC Berkeley Details of the position The CHAI Research Fellowship<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fhumancomp…> Appointment Start Date: Flexible Salary: Negotiated Apply: Via this form<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fforms.gle…> CHAI<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fhumancomp…> has created a new Research Fellowship and is seeking applications for the inaugural Fellows. CHAI Research Fellows will work with CHAI faculty members Stuart Russell, Pieter Abbeel, and Anca Dragan and will also have the opportunity to collaborate with CHAI affiliate faculty<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fhumancomp…> at Berkeley, other faculty members in Berkeley AI Research<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fbair.berk…> (where CHAI is located), and many other institutions. The Fellowship is aimed at training highly qualified postdoctoral researchers to carry out research to advance beneficial AI. Duties of position Given the broad and open-ended mandate of the Center to develop provably beneficial AI systems, the post holder will have considerable freedom to pursue novel research projects in areas such as reasoning, decision making, learning, multi-agent systems, and philosophical foundations using the tools of probability theory, game theory, control theory, etc. Topics of particular current interest include * * * Theoretical foundations of intelligent agent architecture * * * * Expressive formal languages for probability models, including probabilistic programming * * * * Decision making over long time scales * * * * Cumulative lifelong learning * * * * Human-in-the-loop planning and reinforcement learning * * * * Human-robot interaction * * * * The structure and properties of human preferences * * * * Foundations for acting on behalf of multiple humans * * * * Robust cooperation in heterogeneous multiagent systems * * * Mechanism design for human-machine * systems * Depending on their interests, Fellows may also lead research collaborations with other CHAI researchers and Berkeley PhD students, advise more junior student researchers, and teach courses relevant to their research. Candidates for the Fellowship are assessed based on their academic and research achievements, as well as on personal merit and leadership qualities. Qualifications Candidates must have (or be about to obtain) a PhD in a relevant technical discipline (computer science, statistics, mathematics, or theoretical economics) and a record of high-quality published research. A solid understanding of current methods in AI and statistical learning would be an advantage. Candidates need not have done previous work on the AI control problem. Salary & Visa Sponsorship The CHAI Research Fellowships provide funding for one year, renewable up to three years given satisfactory performance. Salary will be commensurate with qualifications and experience and based on UC Berkeley salary scales. The position is eligible for visa sponsorship. CHAI will also contribute to visa costs for Fellows and their dependents (partner and children). Application Apply via this form<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fforms.gle…>. Supporting documents include a resume, a one-page statement of interest describing in general terms the kind of research you would like to undertake at CHAI, and the names and email addresses of two academic referees.

1 0

No AI seminar for the rest of the academic year
by Fuxin Li 09 Jun '21

09 Jun '21

Dear all, There won't be any AI seminar for the rest of the academic year (including summer). Hopefully we will restart in the fall -- maybe with in-person visits but not decided yet. We have had a fantastic lineup of speakers in the past year. Thanks everyone for your attendance in the past academic year and special thanks to Jay Patravali for dealing with the logistics of AI seminar for the whole year! Fuxin -- Fuxin Li Assistant Professor School of Electrical Engineering and Computer Science Oregon State University Corvallis OR 97331 email: lif(a)oregonstate.edu web: http://web.engr.oregonstate.edu/~lif/

1 0

OSU AI Seminar - **Today** 1PM | Jiajun Wu - Stanford University
by Patravali, Jay Sanjay 02 Jun '21

02 Jun '21

Dear All, A gentle reminder for today's AI seminar by Prof. Jiajun Wu from Stanford University. He will be presenting *"*Learning to See the Physical World*"* between* 1.00-2.00PM PT.* Zoom Link: https://tinyurl.com/5wn8zczw Please find attached the poster flyer for more information. *Title*: Learning to See the Physical World *Abstract*: Human intelligence is beyond pattern recognition. From a single image, we're able to explain what we see, reconstruct the scene in 3D, predict what's going to happen, and plan our actions accordingly. In this talk, I will present our recent work on physical scene understanding---building versatile, data-efficient, and generalizable machines that learn to see, reason about, and interact with the physical world. The core idea is to exploit the generic, causal structure behind the world, including knowledge from computer graphics, physics, and language, in the form of approximate simulation engines, and to integrate them with deep learning. Here, deep learning plays two major roles: first, it learns to invert simulation engines for efficient inference; second, it learns to augment simulation engines for constructing powerful forward models. I'll focus on a few topics to demonstrate this idea: building scene representation for both object geometry and physics; learning expressive dynamics models for planning and control; perception and reasoning beyond vision. *Bio*: Jiajun Wu is an Assistant Professor of Computer Science at Stanford University, working on computer vision, machine learning, and computational cognitive science. Before joining Stanford, he was a Visiting Faculty Researcher at Google Research. He received his Ph.D. at the Massachusetts Institute of Technology. Wu's research has been recognized through the ACM Doctoral Dissertation Award Honorable Mention, the AAAI/ACM SIGAI Doctoral Dissertation Award, the 2020 Samsung AI Researcher of the Year, the IROS Best Paper Award on Cognitive Robotics, and faculty research awards and graduate fellowships from Samsung, Amazon, Facebook, Nvidia, and Adobe. Please plan on attending! Thanks, Jay Patravali Graduate Student, CoRIS-EECS. Oregon State University

1 0

AI alignment reading group: Interpretable Machine Learning: Fundamental Principles and 10 Grand Challenges
by Turner, Alex 02 Jun '21

02 Jun '21

This week, we'll be reading Rudin et al's "Interpretable Machine Learning: Fundamental Principles and 10 Grand Challenges": "Interpretability in machine learning (ML) is crucial for high stakes decisions and troubleshooting. In this work, we provide fundamental principles for interpretable ML, and dispel common misunderstandings that dilute the importance of this crucial topic. We also identify 10 technical challenge areas in interpretable machine learning and provide history and background on each problem. Some of these problems are classically important, and some are recent problems that have arisen in the last few years. These problems are: (1) ... (9) Characterization of the "Rashomon set" of good models; and (10) Interpretable reinforcement learning." This is a long paper, but I'm particularly interested in discussing (9): Rashomon sets of models. In what situations should we expect black-box models (e.g. deep neural networks) to have interpretable counterparts (e.g. sparse logical models / decision trees) with similar performance? The answer to this question will help govern competitive pressures for using/not using interpretable models, and also inform the difficulty of supervising the computation performed by potential future human-level ML systems. The paper: https://arxiv.org/abs/2103.11251 We'll meet Friday at 1. https://oregonstate.zoom.us/j/2739792686?pwd=VkRUeHJkYnhvTzlvZzR6YnZWNERKQT… Best, Alex Turner

1 0