Learning Fair Representations
We propose a learning algorithm for fair classification that achieves both group fairness (the proportion of members in a protected group receiving positive classification is identical to the proportion in the population as a whole), and individual fairness (similar individuals should be treated similarly). We formulate fairness as an optimization problem of finding a good representation of the data with two competing goals: to encode the data as well as possible, while simultaneously obfuscating any information about membership in the protected group. We show positive results of our algorithm relative to other known techniques, on three datasets. Moreover, we demonstrate several advantages to our approach. First, our intermediate representation can be used for other classification tasks (i.e., transfer learning is possible); secondly, we take a step toward learning a distance metric which can find important dimensions of the data for classification.
Proxy objectives are a fundamental concept in machine learning. That is, there's a true objective that we care about, but it's hard to compute or estimate, so instead we construct a locally-valid approximation and optimize that. I will examine reinforcement from human feedback with this lens, as a chain of approximations, each of which can widen the gap between the desired and achieved result.
The Societal Impacts of AI
Unleashing the Power of AI-Enabled Digital Twins
Digital twins with Artificial Intelligence (AI) capabilities are computer-generated models of actual products, procedures, or systems that offer real-time performance analysis and insights. The incorporation of AI makes it possible to analyse the data produced by these systems in a more precise and effective manner. These digital twins are employed to simulate, monitor, and improve real systems.
This social event will explore digital twin technologies powered by AI and how they might be utilized to link the virtual and physical worlds. We will give a general overview of AI-enabled digital twin technologies and demonstrate practical applications of AI-enabled digital twins in manufacturing, healthcare, and aerospace. We will also discuss the opportunities and challenges that come with implementing it across different sectors.
There will be group discussions on the challenges and opportunities of digital twins powered by AI at the event to discuss how these technologies will continue to influence the future of various industries.
Target audience: Professionals, researchers, and students who are interested in AI-enabled digital twins and their applications.
Behind the buzzwords. Humans...
We'll have roundtables set up with different themes related to data-centric AI and machine learning. Participants will be welcome to discuss the topic of their choice. Key learnings will be shared at the end of the event.
Early Career Researchers Social
Scrambling to understand your research topic? Pulled several all nighters for your first paper? Wept with joy when you were accepted into your PhD programme? This social is for you! We're here to allow master and 1st-2nd year PhD students chat about their research, getting started, and how to handle it all.
The Role of Generative AI in Shaping the Next Generation of the Metaverse
"This event explores the role of Generative AI in shaping the next generation of the Metaverse, fostering an environment where everyone can be a creator. The program includes presentations from researchers and practitioners from both academia and industry, and will cover:
(1) Recent trends in multimodal content generation research, encompassing the fusion of text, image, and video. The discussion will emphasize maintaining control over the latent space and achieving compositionality in AI-generated content.
(2) The application of neuro-symbolic representations for 3D Generative AI and geometric learning on discrete surfaces in 3D content creation.
(3) Practical implementations of Generative AI within Roblox with live demonstrations of how language models and image generation techniques can streamline the creation of interactive 3D objects and game worlds.
Agenda:
- Mubbasir Kapadia (Roblox and Rutgers): Introduction
- Honglu Zhou (NEC Labs): Illuminating the Metaverse: Unveiling NEC Labs' Journey in Revolutionizing AIGC with Compositionality
- Derek Liu (Roblox Research): Geometric Learning on Discrete Surface Meshes
- Daniel Ritchie (Brown University): Neuro-symbolic Methods for 3D Generative AI
- Kartik Ayyar (Roblox): Generative AI in Action at Roblox
The development and deployment of AI often lacks input from impacted communities such as warehouse workers who directly interact with AI in their jobs. As a two-part social, “Inclusive AI: Magnifying Marginalized Communities” will introduce ways to include and bring in the perspectives of marginalized stakeholder communities. In the first hour, participants will be introduced to case studies that walk through the engagement process with various worker groups. Along with the case studies, participants will reflect on ways to responsibly engage with the broader community in their own work. In the second half, we will host a happy hour for practitioners to connect with others about community engagement, participatory methods, and inclusive AI. Much like speed dating events, we will supply the conversation topics and refreshments; participants bring the curiosity and conversation!