A naukri.com initiative
Medium
4w
396
Image Credit: Medium
Advancing AI research at NAACL 2025
- Capital One researchers and technologists are participating in NAACL 2025 to showcase their latest NLP research and engage with the academic community.
- Their research focuses on multilingual understanding, model efficiency, and AI safety, with presentations, poster sessions, tutorials, and a TrustNLP workshop at Booth 4.
- Research by Capital One Enterprise AI teams includes methods for AI safety, real-world NLP applications, red teaming frameworks, and deployment insights.
- One paper, 'Building Safe GenAI Applications,' provides an overview of red teaming for Large Language Models by Genta Winata, Anirban Das, and Austin Zhang.
- Another collaborative effort introduced 'WorldCuisines,' a benchmark for multilingual visual question answering across diverse cultures, presented by the same authors.
- The 'ProxyLM' paper presents a framework for predicting Language Model performance on multilingual tasks using proxy models, reducing computational overhead.
- Capital One sponsors the TrustNLP workshop, focusing on building trustworthy NLP systems and responsible AI deployment.
- Attendees are invited to connect with Capital One at NAACL 2025 to learn about NLP research, career opportunities, and how AI is transforming the financial industry.
- Opinions expressed in the article are those of individual authors and not necessarily those of Capital One. Trademarks and intellectual property rights belong to their respective owners.
- The article was originally published on the Capital One Tech Medium and discusses ongoing conversations about advancing AI research at NAACL 2025.
Read Full Article
23 Likes
For uninterrupted reading, download the app