Our research at SwAPP lab is primarily focused on the intersection of HPC (parallel computing) and data science (AI) and includes (refer to publication page for the latest papers of our lab):
- Parallelism Discovery, Performance Analysis, and Scheduling
- Smart AutoTuner, profiling, performance analysis, and auto-tuning. [NeurIPS'23], [HPDC'23], [IPDPS'23], [AI4S'22], [IPDPS'22], [TECS'21], [Euro-Par'20],[Euro-Par'19], [ICPP'18], [ICPP'16], [IPDPS'15]
- The parallelization framework DiscoPoP can detect parallelization opportunities and parallel design patterns. [MLSys'23], [IPDPSW'22], [ICS'19], [CPE'19], [ICSE'17], [JSS'16], [TACO'16]
- Correctness and testing techniques for parallel programming. [NEXTA'21], [AIST'21], [IJPP'16] [IJPP'15], [TPDPS'14]
- Resource allocation, scheduling, and AI-assisted methods for HPC clusters and stream data processing platforms. [PDCAT'21], [CCGrid'20], [NCA'19], [PDCAT'19], [JS'15]
- Efficient and Scalable Learning and Inferences
- Accelerating training and inference of deep neural networks (DNNs) with our GNN-RL pipeline using Graph Neural Networks (GNNs), Reinforcement Learning (RL), model compression, and optimization techniques. [SC'22], [ICML'22], [ICCV'21],[arXiv: 2102.03214], [arXiv'20: 2011.12641]
- Boda: Efficiency and productivity framework for deploying deep neural networks across hardware platforms and programming models. [EuroSys'20], [Euro-Par'19 - best paper], [TACO'19], [CF'17], [arXiv'16]
- Federated learning and decentralized learning: [CCGRID'23], [SC'22], [arXiv preprint: 2211], [arXiv:2208.07978], [arXiv: 2106.06921]
- Applied machine learning and edge AI computing: Computer vision and visual computing for different applications such as hydro-ecological models, healthcare, and agricultural applications. [CIS'21], [IJERPH'20], [IRI'20], [arXiv: 2010.04328], [arXiv'19:1909.12217], [Edge AI]
- Software Analytics, Software Engineering Research, and AI for Cyberinfrastructure
- Software engineering research, program analysis, and AI-assisted and analytical methods for parallel programming, HPC and distributed computing, and Cyberinfrastructure. [NeurIPS'23], [MLSys'23], [IPDPS'23], [CIKM'23], [ESEM'22], [AI2ASE'22], [IPDPS'22], [PDP'21], [AIST'21], [LLVM-HPC-SC'20], [RL+SE&PL-FSE'20], [arXiv'19:1907.06205], [arXiv'19:1907.07110], [CPE'18], [ICSE-SEET'17]