Summary of Carin: Constraint-aware and Responsive Inference on Heterogeneous Devices For Single- and Multi-dnn Workloads, by Ioannis Panopoulos et al.
CARIn: Constraint-Aware and Responsive Inference on Heterogeneous Devices for Single- and Multi-DNN Workloads
by Ioannis Panopoulos, Stylianos I. Venieris, Iakovos S. Venieris
First submitted to arxiv on: 2 Sep 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Distributed, Parallel, and Cluster Computing (cs.DC)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper presents CARIn, a novel framework for optimizing the execution of deep neural networks (DNNs) on mobile devices. The framework addresses challenges such as device heterogeneity, multi-DNN execution, and dynamic runtime adaptation. CARIn uses a multi-objective optimization framework and a runtime-aware sorting and search algorithm (RASS) to efficiently adapt to changing conditions and address resource contention issues. The paper evaluates CARIn across various tasks, including text classification, scene recognition, and face analysis, using different model architectures like Convolutional Neural Networks and Transformers. The results show that CARIn achieves a significant enhancement in fair treatment of objectives, outperforming single-model designs by 1.92x and the state-of-the-art OODIn framework by up to 10.69x. Additionally, CARIn eliminates time overhead associated with identifying optimal designs. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about a new way to make deep learning models work better on smartphones. Right now, these models are too big and slow for phones, so people need to find ways to make them faster and more efficient. The authors of this paper created a special tool called CARIn that helps make these models work better on phones. This tool can handle different types of models and even multiple models at the same time. It’s like having a personal trainer for your phone! The researchers tested their tool with lots of different tasks, such as recognizing pictures and understanding text. They found that CARIn makes things run faster and better than before. |
Keywords
* Artificial intelligence * Deep learning * Optimization * Text classification
