# 11-785 Deep Learning https://deeplearning.cs.cmu.edu/shared/project.html [ Back to Course Page](https://deeplearning.cs.cmu.edu/F26/index.html) [Projects](https://deeplearning.cs.cmu.edu/shared/project.html#topics) [Application](https://deeplearning.cs.cmu.edu/shared/project.html#apply) [Legacy Projects](https://deeplearning.cs.cmu.edu/shared/project.html#legacy) **11-785** Introduction to Deep Learning ## Projects Page ## Course Project Guidelines Graduate versions of IDL include a significant weightage on a course project. In 11-785, you are expected to come up with your own project, or use some of the suggested topics below as a starting point. In 11-685, you are expected to work on a guided project, which is a pre-configured project with a set of organized milestones. TAs will help you with both projects: #### 11-785: Open Course Project A primarily student-driven open research project. You may choose your own topic and work in teams of up to 4. * **Core Deliverables:** Midterm Report (20%), Piazza Discussion & Q&A Response (5%), Video Presentation (35%), and Final Report (40%). * **Structure:** Aligned with industry research. Requires periodic self-guided planning, experimentation, writing, and finally peer reviews. * **Theme:** Open-ended topics covering (but not limited to): Vision, NLP, RL, Generative AI, or Computational Biology. #### 11-685: Guided Course Project A mentored and structured project pathway. Ideal for students seeking a more hands-on guidance on executing an end-to-end deep learning project. * **Core Deliverables:** Midterm Report (20%), Piazza Discussion & Q&A Response (5%), Video Presentation (35%), and Final Report (40%). * **Structure:** TAs release pre-configured guided projects with milestone benchmarks and technical constraints. * **Mentorship:** Includes closer, dedicated review sessions and tailored support with starter notebooks, datasets and other resources. **IMPORTANT NOTE ON COMPUTE RESOURCES:** All project related computations MUST be performed on dedicated resources (AWS/GCP/PSC/personal) for model training, testing and inference. You MUST NOT use any resources from the IDL homework allocation on PSC. Guidelines for requesting compute resources for your project can be found in [Recitation 0.13](https://www.youtube.com/watch?v=J4H4bvCLiPw) or Piazza. ## Potential Project Ideas We encourage you to propose a project of your own. But if you need assistance, the project ideas below are intended to provide a starting point. Select the project topic(s) that interests you and your team, and let us know by filling out the form with your preferred project(s). Reinforcement Learning (11-785) ### Reinforcement Learning from Pixels Mentor(s): Kangping Liu This project is on model-based reinforcement learning from the pixel space. It focuses on PlaNet and DreamerV1 on DeepMind Control Suite tasks, including latent dynamics models, replay buffers, planning or actor-critic control. RLWorld ModelsRobotics Coming Soon Coming Soon Coming Soon Select Project AI4Science (11-785) ### Machine Learned Interatomic Potentials Mentor(s): Ron Sarma Build an efficient, accurate and competitive machine learned interatomic potential for molecular systems. There is a significant shift of momemntum in AI4Science, and this project is intended to a step in that direction. MLIPGNNPhysics-informed [ Paper ](https://arxiv.org/abs/2506.23971) [ Dataset ](https://huggingface.co/facebook/OMol25) [ Starter Code ](https://github.com/facebookresearch/fairchem) Select Project Robotics (11-785) ### Atomic skill library for robotic manipulation Mentor(s): Nayesha Gandotra You will pick a robotic manipulation domain/task of choice and set it up in Maniskill. Then, using at least 2 distinct deep learning techniques train a set of skills pertinent to the domain. Performance will be compared across chosen techniques. LLMs Coming Soon Coming Soon Coming Soon Select Project Signal Processing (11-785) ### Hand and Finger Control from Forearm sEMG Mentor(s): Nayesha Gandotra and Max Murphy This project asks: can deep learning tell which attempted hand and finger movements are distinguishable from surface EMG signals recorded at the forearm? And, is it controllable in a graded, proportional way? Perhaps in real time? EMGSignal Processing Coming Soon Private dataset Coming Soon Select Project Perforation and NLP (11-785) ### Perforated Learning for Language Modeling Mentor(s): Rorry Brenner and the Perforated AI Team Perforated Backpropagation is a new deep learning paradigm enabling ML models to be built with more parameter efficiency and data efficiency. This project focuses on perforating various ML models in NLP. NeuroscienceDendritic Optimization [ Paper ](https://arxiv.org/abs/2501.18018) N/A [ Starter Code ](https://github.com/PerforatedAI/PerforatedAI) Select Project Opening Google Form... ## Project Interest Form ### Submit Project Interest Form Use this form to show interest in any project(s) for **11-785** or **11-685**. This is just for book-keeping, and not a formal selection process Go to Application Form #### What You Will Need: * **Student Details:** Full names and Andrew IDs of all team members (up to 4 per team). * **Registered Section:** Specify the section that you are registered for, 11-785 (Open) or 11-685 (Guided). * **Project Interest Statement:** A brief paragraph expresing why you are interested in the project, and what you plan to achieve. * **Core Technical Area:** Select your project's theme (e.g. Vision, NLP, Generative AI, RL, or something else). * **Any Additional Links/Documents (optional):** Relevant GitHub repositories, pre-existing research references, or personal portfolios. * **Compute Requests (optional):** Details if you are requesting additional AWS, GCP, or PSC resources. ## Legacy Projects Explore a curated list of student projects from previous semesters of IDL. They serve as a benchmark of the expected scope, quality, and scientific rigor. Spring 2020 #### Spring 2020 Project Archive Browse the archived top 25 projects and their submitted reports. [ Project Reports](https://deeplearning.cs.cmu.edu/shared/projects/S20/gallery.html) [ Project Videos](https://www.youtube.com/playlist?list=PLp-0K3kfddPyyQzmID534-R1CnGJs0Ls-) Fall 2020 #### Fall 2020 Project Archive Browse the complete Fall 2020 collection of projects and reports. [ Project Reports](https://deeplearning.cs.cmu.edu/shared/projects/F20/gallery.html) [ Project Videos](https://www.youtube.com/playlist?list=PLp-0K3kfddPzcVQIc7eh0kOaOLhtWksJo) Spring 2024 #### Spring 2024 Project Archive Browse all Spring 2024 project submissions and reports. [ Project Reports](https://deeplearning.cs.cmu.edu/shared/projects/S24/gallery.html) Fall 2024 #### Fall 2024 Project Archive Browse all Fall 2024 project submissions and reports. [ Project Reports](https://deeplearning.cs.cmu.edu/shared/projects/F24/gallery.html) Spring 2025 #### Spring 2025 Project Archive Browse all Spring 2025 project submissions and reports. [ Project Reports](https://deeplearning.cs.cmu.edu/shared/projects/S25/gallery.html) Fall 2025 #### Fall 2025 Project Archive Browse all Fall 2025 project submissions and reports. [ Project Reports](https://deeplearning.cs.cmu.edu/shared/projects/F25/gallery.html) Spring 2026 #### Spring 2026 Project Archive Browse all Spring 2026 project submissions and reports. [ Project Reports](https://deeplearning.cs.cmu.edu/shared/projects/S26/gallery.html) **Supported by** ![Google](https://deeplearning.cs.cmu.edu/F26/images/google-logo-transparent.png) 11-785 Introduction to Deep Learning | Carnegie Mellon University