MBZUAI Skyrockets In Global AI Rankings With Groundbreaking Research In 2024

Mohamed bin Zayed University of Artificial Intelligence (MBZUAI), the first graduate-level, research-based AI university globally, continues to expand its publication of innovative AI research. In the first half of 2024, MBZUAI's community, comprising over 80 faculty members, more than 200 researchers, and numerous students, published over 300 papers at top-tier AI conferences. This included 39 papers at the International Conference on Learning Representations (ICLR) in May.

In 2023, MBZUAI achieved significant success with 612 papers published at leading venues. Highlights included 30 papers at the International Conference on Computer Vision (ICCV), 34 at the IEEE / CVF Computer Vision and Pattern Recognition Conference (CVPR), 44 at Empirical Methods in Natural Language Processing (EMNLP), and 53 at the Conference on Neural Information Processing Systems (NeurIPS).

Focus on AI Research

Five years since its establishment, MBZUAI is now ranked among the world's top 100 universities in computer science. It holds a top-20 position globally in AI, computer vision, machine learning, natural language processing (NLP), and robotics according to CSRankings.

One notable research paper from MBZUAI addresses the misuse of text generated by large language models (LLMs). Collaborating with international partners, MBZUAI researchers developed resources for detecting LLM-generated text across multiple languages and domains. The paper titled 'M4: Multi-generator, multi-domain, and multi-lingual black-box machine-generated text detection’ won the Best Resource Paper Award at the European Chapter of the Association for Computational Linguistics Conference 2024 (EACL) in March.

Advancements in Genetic Research

Professor Kun Zhang from MBZUAI collaborated with Ph.D. student Gongxu Luo and researchers from American universities to enhance gene-sequencing analysis. Their model improves understanding of diseases like cancer by addressing dropouts in single-cell RNA sequencing data. The paper 'Gene Regulatory Network Inference In The Presence Of Dropouts: A Causal View’ was presented at ICLR and marks a significant advancement in genetic research.

Another breakthrough involves an algorithm developed by William de Vazelhes, one of MBZUAI's first Ph.D. graduates. Alongside faculty and a fellow Ph.D. student, he created a method that enhances training efficiency for models using hard-thresholding techniques. The paper 'New Insight of Variance reduce in Zero-Order Hard-Thresholding: Mitigating Gradient Error and Expansivity Contradictions’ was published at ICLR 2024.

Innovative Vision Language Model

A team from MBZUAI led the development of Grounding Large Multimodal Model (GLaMM), an advanced vision language model that enables detailed visual understanding through pixel-level interaction between text and images. The paper 'GLaMM: Pixel grounding large multimodal model’ was presented at CVPR 2024 in Seattle. GLaMM has already garnered over 50 citations and 600 stars on GitHub.

Dr. Xiaodan Liang and Professor Xiaojun Chang from MBZUAI's Computer Vision Department collaborated internationally to improve AI vision transformers' efficiency. Their research demonstrated that replacing certain transformer layers with simpler multilayer perceptron (MLP) layers could maintain performance while reducing model size. The paper 'MLP can be a good Transformer Learner’ was presented as an oral presentation at CVPR 2024 and was nominated for a Best Paper Award.

The continued success of MBZUAI highlights its role as a leading institution in AI research, contributing significantly to advancements across various fields including genetics, machine learning algorithms, and vision-language models.