Vision Language Models - Connect Session

ResearchTrend.AI VLM Connect Session: Cultural Reasoning & Test-Time Augmentation!

We are excited to announce our upcoming biweekly Vision-Language Model (VLM) Connect Session on ResearchTrend.AI!

This virtual session 💻 features two groundbreaking presentations from leading researchers 🧑‍🔬, diving into the critical areas of VLM evaluation and generation enhancement.

Agenda (UTC) - Monday, December 8th

07:00 - 07:30: Burak Satar, PhD (Singapore Management University)
📄 Paper: Seeing Culture: A Benchmark for Visual Reasoning and Grounding
💡 Abstract: VLMs often fall short in cultural reasoning, especially in underrepresented cultures. Burak will introduce the Seeing Culture Benchmark (SCB), which focuses on Southeast Asian countries. The SCB is a novel, two-stage evaluation: 1) VQA to select the correct cultural option, followed by 2) Segmentation of the relevant artifact as evidence of reasoning. This highlights the crucial disparity between visual reasoning and spatial grounding in culturally nuanced scenarios.

07:30 - 08:00: Yapeng Mi
📄 Paper: MILR: Improving Multimodal Image Generation via Test-Time Latent Reasoning
💡 Abstract: Existing reasoning methods for image generation rely on fine-tuning or limit reasoning to a single modality. Yapeng will present MILR (Multimodal Image generation via test-time Latent Reasoning), a novel method that jointly reasons over image and text in a unified latent vector space—all at test time. Implemented via policy gradient guided by an image quality critic, MILR achieves state-of-the-art results on multiple benchmarks, particularly showing impressive gains (80% improvement on the knowledge-intensive WISE benchmark) in temporal and cultural reasoning tasks.

🌟 This is a fantastic opportunity to engage directly with research that addresses cultural competence and inference-time optimization in multimodal AI.

🗓️ Time: 7:00 AM - 8:00 AM UTC 📍 Location: Virtual

👉 Register for this event here: https://lnkd.in/ehSQ9Gvc

Don't miss our future sessions! 📆 Find out more about upcoming events: https://lnkd.in/g7-iczUp