AFMAML 2027 welcomes relevant paper submissions from researchers in academia, industry, and government, such as students, engineers, practitioners, scientists, and policy makers. We welcome paper submissions with original technical and scientific research results in relevant topics.
Main Topics of Interest:
Track 1: Foundation Model Architectures and Theories |
|---|
| Novel neural network architectures for foundation models (transformers, state-space models, diffusion models, etc.) |
| Scaling laws and emergent behaviors in large-scale models |
| Theoretical foundations of pre-training and representation learning |
| Architectural innovations for efficiency, interpretability, and generalization |
| Multi-modal foundation model architectures (vision-language, audio-visual, etc.) |
| Sparse models, mixture-of-experts, and modular architectures |
| Foundational model design for scientific computing and domain-specific applications |
Track 2: Training, Fine-Tuning, and Adaptation Methods |
| Large-scale pre-training algorithms and distributed training strategies |
| Supervised fine-tuning, instruction tuning, and model alignment |
| Parameter-efficient fine-tuning (LoRA, adapters, prefix-tuning, etc.) |
| Reinforcement learning from human feedback (RLHF) and AI feedback (RLAIF) |
| Transfer learning, few-shot learning, and meta-learning |
| Continual learning and adaptation to dynamic environments |
| Model compression, distillation, and quantization for efficient deployment |
Track 3: Applied Machine Learning with Foundation Models |
| Large language models (LLMs) for natural language processing and generation |
| Multi-modal learning and cross-modal understanding |
| Retrieval-augmented generation (RAG) and knowledge-enhanced models |
| Agentic AI and autonomous systems powered by foundation models |
| Foundation models for computer vision, speech, and audio processing |
| Generative AI: diffusion models, GANs, VAEs, and beyond |
| Foundation models for structured and unstructured data analytics |
| Domain-specific foundation models (healthcare, finance, law, education, etc.) |
Track 4: Trustworthy, Efficient, and Scalable Foundation Models |
| Explainability, interpretability, and transparency in foundation models |
| Fairness, bias detection, and mitigation in large-scale AI systems |
| Privacy-preserving machine learning and federated learning |
| Robustness, adversarial defense, and model safety |
| Energy-efficient training and inference for sustainable AI |
| Benchmarking, evaluation, and responsible deployment of foundation models |
| Regulatory compliance and ethical AI governance |
Track 5: Emerging Applications and Interdisciplinary Innovations |
| Foundation models for scientific discovery (physics, chemistry, biology, climate science) |
| AI for healthcare, biomedical informatics, and medical imaging |
| Autonomous driving and robotics empowered by foundation models |
| Foundation models for cybersecurity, IoT, and edge intelligence |
| Software engineering and code generation with foundation models |
| AI for social good, education, and creative industries |
| Human-AI collaboration and interactive AI systems |
