Select a calendar:
Filter November Events by Event Type:
SUNMONTUEWEDTHUFRISAT
Events for November 16, 2023
-
NL Seminar- Cultural Knowledge and Cultural Biases: Analyzing the Multilingual Performance of Text-to-Image Models
Thu, Nov 16, 2023 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Michael Saxon, UCSB
Talk Title: Cultural Knowledge and Cultural Biases: Analyzing the Multilingual Performance of Text-to-Image Models
Abstract: REMINDER: Meeting hosts only admit guests that they know to the Zoom meeting. Hence, you are highly encouraged to use your USC account to sign into Zoom. Meeting hosts only admit guests that they know to the Zoom meeting. Hence, you’re highly encouraged to use your USC account to sign into Zoom. If you’re an outside visitor, please provide your: Full Name, Title and Name of Workplace to (nlg-seminar-host(at)isi.edu) beforehand so we’ll be aware of your attendance. Also, let us know if you plan to attend in-person or virtually. More Info for NL Seminars can be found at: https://nlg.isi.edu/nl-seminar/ Despite being ostensibly trained on solely English data, most text-to-image (T2I) models carry some degree of multilingual capability, with significant variation in performance between models and languages. To guide the future development of T2I systems, both measuring and qualitatively analyzing these language-specific performance variations is desirable, to mitigate cross-lingual disparities in performance as well as language-specific demographic biases.To quantify multilingual performance we introduce the Conceptual Coverage Across Languages (CoCo-CroLa) benchmark, which allows us to measure the "possession" of a set of tangible noun "concepts" across English, Spanish, German, Chinese, Japanese, Hebrew, and Indonesian. This technique allows us to estimate how well-suited a model is to a target language as well as identify model-specific weaknesses, spurious correlations, and biases without any a-priori assumptions of their form. We demonstrate how it can be used to rank T2I models in terms of multilinguality, and that despite its simplicity our method captures the necessary conditions for the impressive “creative” generative abilities users expect from T2I models.We then build on this benchmarking work with a detailed qualitative analysis of “failure” and “success” cases for specific concepts. Even in the “possession” case, concepts are expressed differently across languages. These qualitative cross-lingual variations in model behaviors form a continuous spectrum of ethical acceptability, running the gamut from culturally variable popular dog breeds to racially-biased sexualization in depictions of women. While the edge cases are easy to laud or condemn, drawing the line of acceptability in between them is an open ethical question as well as an open technical challenge. Unfortunately, interventions that successfully remove the most deleterious biases also erase cultural distinctiveness, motivating a need for more targeted interventions in future work.
Biography: Michael Saxon is a CS Ph.D. candidate in the NLP Group at the University of California, Santa Barbara. His research is driven by a desire to improve our objective understanding of the semantic capabilities of large generative AI systems, in particular generative image and language models. Toward this goal he focuses on developing novel data resources and metrics for to model semantic phenomena in generative model, as well as techniques for model-driven dataset improvement to remove biases and spurious correlations. He has previously interned at Meta AI and Amazon working on NLP and speech, and is supported by the NSF Graduate Research Fellowship Program.
Host: Jon May and Justin Cho
More Info: https://nlg.isi.edu/nl-seminar/
Webcast: https://youtu.be/nlu57ZSKbi0Location: Information Science Institute (ISI) - Virtual and ISI-Conf Rm#689
WebCast Link: https://youtu.be/nlu57ZSKbi0
Audiences: Everyone Is Invited
Contact: Pete Zamar
Event Link: https://nlg.isi.edu/nl-seminar/
-
Trojan Talk: ARUP DEI Virtual Info Session
Thu, Nov 16, 2023 @ 12:00 PM - 02:00 PM
Viterbi School of Engineering Career Connections
Workshops & Infosessions
Arup is hiring! Join us on Brazen to speak with our Campus Recruiters and Arup representatives apart of our Employee Resource Groups. Learn more about our culture, our commitment to equity, diversity, and inclusion, and hear about current graduate and internship opportunities. Bring your questions!
Time: Thursday, November 16th, 12-2 pm PST
Location: Virtual on Brazen - Click here to register for this event.
External employer-hosted events and activities are not affiliated with the USC Viterbi Career Connections Office. They are posted on Viterbi Career Connections because they may be of interest to members of the Viterbi community. The inclusion of any activity does not indicate USC sponsorship or endorsement of that activity or event. It is the responsibility of the participant to apply due diligence, exercise caution when participating, and report concerns to vcareers@usc.edu" target="_blank" rel="noreferrer noopener">vcareers@usc.eduLocation: Virtual
Audiences: Everyone Is Invited
Contact: RTH 218 Viterbi Career Connections
-
PhD Thesis Defense - Iordanis Fostiropoulos
Thu, Nov 16, 2023 @ 01:00 PM - 02:00 PM
Thomas Lord Department of Computer Science
University Calendar
PhD Thesis Defense - Iordanis Fostiropoulos
Committee Members: Laurent Itti, Mohammad Soleymani, Stefanos Nikolaidis, Nicolas Schweighofer
Title: Towards Efficient Task Generalization
Abstract: Current practices in Machine Learning (ML) require a model to be trained iteratively on novel examples and tasks. The same model generalizes poorly on previously learned data, where we empirically observe 'Catastrophic Forgetting'. Generalizing across tasks can be trivially solved when there is no restriction on the computational resources. We find that current state-of-the-art fails catastrophically to perform robustly when presented with a large sequence of tasks with large domain gaps. Additionally, simpler methods have improved generalization compared to state-of-the-art methods. While current methods suffer in computational performance. In this talk, we present our work that introduces a framework for efficiently learning a large sequence of tasks by utilizing several experts under strict computational constraints. Last, we discuss future improvements of our method and industrial applications, for example, to self-driving carsLocation: Hughes Aircraft Electrical Engineering Center (EEB) - 110
Audiences: Everyone Is Invited
Contact: Melissa Ochoa
-
CAIS Webinar: Sidestepping the Black-Box: A New Paradigm for Explainable AI
Thu, Nov 16, 2023 @ 02:00 PM - 03:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Dr. Amulya Yadav, PNC Technologies Career Development Assistant Professor (Penn State University)
Talk Title: Sidestepping the Black-Box: A New Paradigm for Explainable AI
Abstract: Existing work in Explainable Artificial Intelligence (XAI) has been focused on developing techniques to interpret decisions made by pre trained and black box machine learning (ML) models. This black box assumption is reasonable in a lot of settings, e.g., explaining Amazons recommender systems requires assuming a black box model because it is infeasible to assume glass box access to Amazons proprietary models, etc. However, I argue that in many real world settings (especially those that pertain to low resource domains), the black box assumption is unnecessary, undesirable, and often, overly limiting. In this talk, I motivate the need to move away from the black box assumption of XAI by discussing two deployed use cases of responsible AI research i. automated tele triage for poor pregnant women in Kenya, and ii. raising awareness of HIV among homeless youth in Los Angeles. Through my experiences with the deployment of AI in these domains, we will argue the need for a new paradigm in explainable AI. Next, I will discuss two new frameworks i. CounterNet, a novel end to end learning framework which integrates Machine Learning (ML) model training and the generation of corresponding counterfactual (CF) explanations into a single end to end pipeline and ii. RoCourseNet, a training framework that jointly optimizes predictions and recourses that are robust to future data shifts.
This lecture satisfies requirements for CSCI 591: Research Colloquium
Biography: Amulya Yadav is the PNC Technologies Career Development Assistant Professor in the College of Information Sciences and Technology at Penn State University, where he serves as Director of the RAISE Research Lab. He is also the Associate Director (Programs) at the Center for Socially Responsible AI at Penn State. Amulyas research work in the field of Responsible AI and Artificial Intelligence for Social Good focuses on developing theoretically grounded approaches to real world problems that can have an impact in the field. His algorithms have been deployed in the real world, particularly in the field of public health and wildlife protection. Amulya is a recipient of the AAMAS 2016 Best Student Paper Award, the AAAI 2017 Best Video and Best Student Video Award, the IDEAS 2016 Most Visionary Paper Award, and the AAMAS 2017 Best Paper Award nomination. His work has also been highlighted by Mashable.com as one of 26 incredible innovations that improved the world in 2015.
Amulya holds a Ph.D. in Computer Science from the University of Southern California, and a B. Tech. in Computer Science and Engineering from Indian Institute of Technology (IIT), Patna.
Register for the Zoom webinar here: https://usc.zoom.us/webinar/register/WN_nPykyeAAQH-B3R6p5-kezg
Host: CAIS
More Info: https://usc.zoom.us/webinar/register/WN_nPykyeAAQH-B3R6p5-kezg
Audiences: Everyone Is Invited
Contact: Melissa Ochoa
Event Link: https://usc.zoom.us/webinar/register/WN_nPykyeAAQH-B3R6p5-kezg
-
Munushian Distinguished Lecture - Eli Yablonovitch, Thursday, Nov. 16th at 2pm in EEB 132
Thu, Nov 16, 2023 @ 02:00 PM - 03:30 PM
Ming Hsieh Department of Electrical and Computer Engineering
Conferences, Lectures, & Seminars
Speaker: Eli Yablonovitch, EECS - University of California, Berkeley
Talk Title: Physics does Optimization (for Free); A New Approach Toward Computation
Series: Munushian Visiting Seminar Series
Abstract: Optimization is vital to science, engineering, and artificial intelligence. It is usually done digitally, but every physics inequality performs optimization in the normal course of dynamical evolution-for free. In driven systems we have Onsager's principle of minimum heat generation. Physics-based optimization usually relies upon this inequality. Optical Onsager machines can run 10^7 times faster than conventional machines, while consuming far less power.
Biography: Prof. Yablonovitch introduced the idea that strained semiconductor lasers could have superior performance due to reduced valence band (hole) effective mass. With almost every human interaction with the internet, optical telecommunication occurs by strained semiconductor lasers.
He is regarded as a Father of the Photonic BandGap concept, and he coined the term "Photonic Crystal". The geometrical structure of the first experimentally realized Photonic bandgap, is sometimes called "Yablonovite".
In his photovoltaic research, Yablonovitch introduced the 4(n squared) ("Yablonovitch Limit") light-trapping factor that is in worldwide use, for almost all commercial solar panels.
His mantra that "a great solar cell also needs to be a great LED", is the basis of the world record solar cells: single-junction 29.1% efficiency; dual-junction 31.5%; quadruple-junction 38.8% efficiency; all at 1 sun.
His cellphone antenna company, Ethertronics Inc., shipped over 2x10^9 antennas.He was also a co-Founder of Luxtera Inc., the pioneer in Silicon Photonics, now part of Cisco.
He co-Founded Luminescent Inc., the company that originated "Inverse Lithography Technology".
Host: ECE-Electrophysics
Webcast: Meeting ID: 96220203431 Pass Code: 949129More Information: Eli Yablonovitch Flyer.pdf
Location: Hughes Aircraft Electrical Engineering Center (EEB) - 132
WebCast Link: Meeting ID: 96220203431 Pass Code: 949129
Audiences: Everyone Is Invited
Contact: Marilyn Poplawski
-
2023 Eberhardt Rechtin Keynote Lecture
Thu, Nov 16, 2023 @ 04:00 PM - 06:00 PM
Daniel J. Epstein Department of Industrial and Systems Engineering
Conferences, Lectures, & Seminars
Speaker: Dr. Mark S. Daskin, Clyde W. Johnson Collegiate Professorship, Emeritus; Immediate past Department Chair of the Industrial and Operations Engineering Department at the University of Michigan
Talk Title: Core Principles of Operations Management
Host: Epstein ISE Dept.
More Info: ***Please send email to: owh@usc.edu to RSVP***
More Information: 2023 Recthin Lecture flyer.jpg
Location: USC Hotel, Center Ballroom
Audiences: Everyone Is Invited
Contact: Grace Owh
Event Link: ***Please send email to: owh@usc.edu to RSVP***
-
WIE Meets WII
Thu, Nov 16, 2023 @ 05:30 PM - 07:30 PM
USC Viterbi School of Engineering
Conferences, Lectures, & Seminars
The event was created to help our fellow female engineering students gain perspective on what it is like to work in the industry and get advice on transitioning from college to the workplace. Our theme this year is âIf you can Dream it, you can Be itâ.
Filming Notice:
The University of Southern California is photographing and or video recording the event in which you are participating and or attending. By your presence in this area, you acknowledge that you have been informed that you may be photographed and/or recorded as part of the program/event.
Feel free to contact tfederic@usc.edu if you have any questions.
Location: Sign into EngageSC to View Location
Audiences: Everyone Is Invited
Contact: Thelma Federico Zaragoza
Event Link: https://engage.usc.edu/WIE/rsvp?id=393327
