AI BriefingLG AI Research 379
Key point
Introduces five major AI research trends and key papers in CV, LLM, and more that drew attention at NeurIPS 2023.
Details
We take a look at the major research trends from NeurIPS 2023, the world's most prestigious AI conference. This conference featured five key topics: Computer Vision (CV), Large Language Model (LLM), Large Multimodal Model (LMM), Diffusion Model, and Reinforcement Learning (RL).
In the Computer Vision (CV) field, the evolution toward Vision Foundation Models that handle various output types stood out. As a representative study, Hummingbird proposed a Self Supervised Learning method using a Memory Bank to improve Dense output performance, while SiamMAE demonstrated that effective learning is also possible in the video domain through a Siamese Encoder.
In the Large Language Model (LLM) field, attempts continued to overcome existing models' limitations such as Hallucination, lack of mathematical reasoning ability, and reduced ability to acquire up-to-date information. In particular, Toolformer drew attention by presenting a method that expands the model's capabilities by having it use external Tool APIs on its own.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.