Understanding Localization by a Tailored GPT
Open Access
- 3 June 2024
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 318-330
Abstract
Conventional deep learning approaches for indoor localization often suffer from their reliance on high-quality training samples and display limited adaptability across varied scenarios. To address these challenges, we repurpose the Transformer model, celebrated for its profound contextual insights, to explore the underlying principles of indoor localization. Our microbenchmark results compellingly demonstrate the superiority of our approach, showing improvements of 30% to 70% across a diverse set of 50 scenarios compared to other state-of-the-art methods. In conclusion, we propose a specialized Generative Pre-training Transformer (GPT) variant, termed LocGPT, configured with 36 million parameters that are tailored to facilitate transfer learning. By fine-tuning this pre-trained model, we achieve near-par accuracy using merely half the conventional dataset, thereby heralding a pioneering stride in transfer learning within the indoor localization domain.Keywords
Funding Information
- UGC/GRF (15204820, 15215421)
- NSFC Key Program (61932017)
- Innovation and Technology Fund (ITS/099/21)
- Shenzhen Municipal Fundamental Research Program (JCYJ20230807140410022)
This publication has 30 references indexed in Scilit:
- Supervised and semi-supervised deep probabilistic models for indoor positioning problemsNeurocomputing, 2021
- Radio ranging with ultrahigh resolution using a harmonic radio-frequency identification systemNature Electronics, 2019
- Deep Convolutional Neural Networks for Indoor Localization with CSI ImagesIEEE Transactions on Network Science and Engineering, 2018
- A Novel Phase-Difference Fingerprinting Technique for Localization of Unknown EmittersIEEE Transactions on Vehicular Technology, 2017
- SpotFiACM SIGCOMM Computer Communication Review, 2015
- Accurate Indoor Ranging by Broadband Harmonic Generation in Passive NLTL Backscatter TagsIEEE Transactions on Microwave Theory and Techniques, 2014
- Increased Range Bistatic Scatter RadioIEEE Transactions on Communications, 2014
- From RSSI to CSIACM Computing Surveys, 2013
- Dude, where's my card?ACM SIGCOMM Computer Communication Review, 2013
- The Horus WLAN location determination system2005