..
CS6316 Project Ideas
Instructions
- Improve existing algorithms for a well-studied machine learning problem to address their shortcomings
- Investigate a new learning problem that is seldom studied before
- Refer to ICML, NeurIPS, ICLR, AAAI, IJCAI, JMLR, TPAMI
Recent Hot topics
ICML
- Graph
- Chemistry/Protein
- Audio/Speech
- diffusion
- Multimodal
- Audio/Speech
- LLM
- Watermark LLM
- software/code
- Federated Learning
- Reinforcement Learning
- Bandits
- Causal/Treatment
- Timeseries
- Spike NN
NeurIPS
- Applications
- Chemistry and Drug Discovery
- Climate
- Everything Else
- Genetics, Cell Biology, etc
- Health
- Language, Speech and Dialog
- Neuroscience, Cognitive Science
- Physics
- Robotics
- Time Series
- Computer Vision
- Rendering &; Reconstruction
- Classification and Understanding
- Everything Else
- Image and Video Generation
- Segmentation
- Vision Models & Multimodal
- Data-centric AI
- Data augmentation
- Data-centric AI methods and tools
- Everything Else
- Deep Learning
- Algorithms
- Attention Mechanisms
- Everything Else
- Generative Models and Autoencoders
- Graph Neural Networks
- Robustness
- Theory
- General Machine Learning
- Causality
- Everything Else
- Representation Learning
- Transfer, Multitask and Meta-learning
- Optimization
- Bayesian & Black-box Optimization
- Discrete Optimization
- Everything Else
- Large Scale, Parallel and Distributed
- Neural Network Optimization
- Optimization and Learning under Uncertainty
- Stochastic & Convex Optimization
- Probabilistic Methods
- Bayesian Models and Methods
- Everything Else
- Gaussian Processes
- Monte Carlo and Sampling Methods
- Reinforcement Learning
- Batch Offline
- Deep RL
- Everything Else
- Multi-agent
- Social Aspects
- Accountability, Transparency and Interpretability
- Everything Else
- Fairness, Equity, Justice and Safety
- Privacy-preserving Statistics and Machine Learning
- Trustworthy Machine Learning
- Theory
- Domain Adaptation and Transfer Learning
- Everything Else
- Game Theory
- Learning Theory
- Reinforcement Learning and Planning
Ideas
- Any way to integrate LoRA for fine tuning in a novel setting
LoRA For continued pre-training