- Diversity, Equity, and Inclusion
- Research News
- Department Life
- Oral History of Cornell CS
- Department Timeline
- Job Postings
- Ithaca Info
- Internal info
- Graduation Information
- Cornell Tech Colloquium
- Student Colloquium
- CS Colloquium
- Conway-Walker Lecture Series
- Salton Lecture Series
- Seminars / Lectures
- Big Red Hacks
- Cornell University High School Programming Contests 2023
- Game Design Initiative
- CSMore: The Rising Sophomore Summer Program in Computer Science
- Explore CS Research
- Research Night
Speaker 1: Yiqing Hua
Title: Characterizing and Mitigating Threats to Trust and Safety Online
Abstract: Online trust and safety is constantly threatened by abusive behaviors that cause real human harm. Among the numerous threats, online harassment suppresses voices, and misleading information and propaganda undermine public trust. Existing methods to combat these threats are often not sufficient, as adversaries may abuse and exploit technologies in nuanced ways, and mitigation strategies don't always reflect users' needs. In this talk, I will present my work on characterizing threats, and empowering users with new techniques to combat these threats. First, I will discuss the challenges and solutions to surface subtle and nuanced threats from a large corpus. Second, I will demonstrate the importance of characterizing user participation in adversarial activities, to inform better moderation policy design. Lastly, I will introduce my work on developing privacy-preserving abuse mitigation techniques, to allow user-side warnings of misinformation images in end-to-end encrypted environments.
Bio: Yiqing Hua is a PhD candidate in Computer Science at Cornell Tech, Cornell University. Her research lies in the intersection of social computing and security and privacy. Her work focuses on characterizing threats to online trust and safety, and enabling abuse mitigation in privacy-sensitive environments.
Speaker 2: Alane Suhr
Title: Language and Learning in Collaborative Interactions
Abstract: Systems that use natural language with human collaborators enable non-experts to access complex systems like robots and databases. The challenges and opportunities presented by such interactive, language-using systems have yet to be thoroughly explored. For example, systems must be able to reason about how language meaning depends on interaction context, including the world around them and the interaction history. Interactions with human users, which are full of implicit and explicit feedback about system behavior, provide a new opportunity for learning through interaction. I will first describe my two central research goals: building systems that ground language in context, and developing learning methods that train models during interaction. I will then introduce CerealBar, a platform that we designed to study challenges of learning and using language in collaborative interactions. Finally, I will highlight how we train an instruction-following system to be robust to error propagation, which is a common problem that arises in interactive scenarios.
Bio: Alane Suhr is a final-year PhD student at Cornell University (based in NYC at Cornell Tech) advised by Yoav Artzi. Alane’s interests include language grounding in context, and learning and using language in interaction.
Speaker 3: Mengqi (Mandy) Xia
Title: Physically Realistic Rendering of Complex Materials Using Wave Optics and an Integrated Perspective
Abstract: Compelling visual effects and immersive virtual environments demand realistic material models that describe how light interacts with them. Important appearance effects are missing in conventional material models due to the built-in ray optics assumption. For example, rendered backlit hair is much dimmer than it is in real life and colorful glints are missing in dark animal fur. Moreover, materials can undergo physical and chemical processes that alter their optical properties and in turn influence how we perceive them. Cooking is an example process where material changes are challenging to simulate and render. To improve the realism of visual effects and VR/AR applications, advanced material models that consider the wave nature of light and describe the time-evolution of materials are essential.
In this talk, I will focus on our work that applies wave optics to more realistically render human hair, animal fur and cloth fibers.We propose the first wave optics based fiber scattering model, introducing an azimuthal scattering function that comes from a full-wave simulation. The simulation models light interaction with a fiber by solving Maxwell's equations using the boundary element method (BEM). I will also briefly mention our ongoing work that combines simulation of fluid dynamics and physical optics to reproduce iridescent water droplets from a cup of hot tea. Following this example, I will describe my vision for an integrated study of appearance and dynamics of complex materials, which can significantly broaden the range of remarkable effects we can reproduce and further the potential of mixed reality technologies.
Bio: Mandy Xia is a Ph.D. candidate in Computer Science at Cornell University where she is advised by Prof. Steve Marschner. She works on physically-based rendering and studies how light interacts with complex materials. She obtained her bachelor's degree in Applied Mathematics from University of California, Los Angeles (UCLA) in 2016.