Machine Learning Lab

MLL

Sharif University of Technology



What is MLL Lab?

The Machine Learning Lab (MLL), under the supervision of Dr. Soleymani, is a cutting-edge research center based in Sharif University of Technology, Tehran, Iran. MLL is dedicated to exploring a wide range of critical topics in the field of machine learning, from generalization to compositionality.

Research Area

  • Generalization
  • Compositional Learning
  • Reinforcement Learning
  • Generative Models
  • Vision-language Models

Contact Info

People

Dr. Mahdieh Soleymani

Associate Professor

Email: soleymani@sharif.edu

Google Scholar: Dr-Soleymani

GitHub: -

LinkedIn: -

Hosein Hasani

PhD Student

Email: hosein.hasani@sharif.edu

Google Scholar: Hosein-Hasani

GitHub: -

LinkedIn: -

Mohammad Mahdi Samiei

PhD Student

Email: mm.samiei@sharif.edu

Google Scholar: -

GitHub: -

LinkedIn: -

Negin Hashemi

PhD Student

Email: -

Google Scholar: -

GitHub: -

LinkedIn: -

Mahdi Ghaznavi

MS Student

Email: -

Google Scholar: -

GitHub: -

LinkedIn: -

Soroush Vafaie Tabar

MS Student

Email: Soroush.vafaie96@sharif.edu

Google Scholar: -

GitHub: -

LinkedIn: -

Ali Abdollahi

MS Student

Email: ali.abdollahi024@student.sharif.edu

Google Scholar: -

GitHub: -

LinkedIn: ali-abdollahi024a

Faridoun Mehri

MS Student

Email: f.meh16@student.sharif.edu

Google Scholar: -

GitHub: NightMachinery

LinkedIn: feraidoon-mehri

Ali Bababeyg

MS Student

Email: ali.bababeig@ce.sharif.edu

Google Scholar: -

GitHub: -

LinkedIn: -

Hadi Hoseini

MS Student

Email: -

Google Scholar: -

GitHub: -

LinkedIn: -

Yousef Javaherian

MS Student

Email: -

Google Scholar: -

GitHub: -

LinkedIn: -

Courses

Artificial intelligence

Fall 2023

Website: ...

Large Language Models

Fall 2023

Website: ...

Moder Information Retrieval

Spring 2024

Website: ...

Deep Learning

Spring 2024

Website: ...

Projects

Compositional Generation

In recent years, text-to-image (T2I) diffusion models such as Stable Diffusion and DALL-E have shown promising performance in generating realistic, creative, diverse, and high-quality images from textual descriptions. These models leverage the iterative denoising process and text embeddings through the cross-attention mechanism to generate images used in many applications across various domains. However, despite their impressive capabilities, these models often struggle to faithfully capture all the entities, attributes, and relationships described in the input prompt, leading to various compositional misalignments such as entity missing, improper attribute binding, wrong spatial relationships, and counting problems. We aim to add compositional generation capability into T2I models to overcome the mentioned compositional generation failure modes.

Spurious Correlation

Description

Open Positions

Spurious Correlation

Requirements: ...

Publications

Divide and Conquer: Two-Level Problem Remodeling for Large-Scale Few-Shot Learning

R0-FoMo - 2023

Autors

Description

Link

Send a message