DOI QR코드

DOI QR Code

A Study of Unified Framework with Light Weight Artificial Intelligence Hardware for Broad range of Applications

다중 애플리케이션 처리를 위한 경량 인공지능 하드웨어 기반 통합 프레임워크 연구

  • Received : 2019.09.06
  • Accepted : 2019.10.15
  • Published : 2019.10.31

Abstract

A lightweight artificial intelligence hardware has made great strides in many application areas. In general, a lightweight artificial intelligence system consist of lightweight artificial intelligence engine and preprocessor including feature selection, generation, extraction, and normalization. In order to achieve optimal performance in broad range of applications, lightweight artificial intelligence system needs to choose a good preprocessing function and set their respective hyper-parameters. This paper proposes a unified framework for a lightweight artificial intelligence system and utilization method for finding models with optimal performance to use on a given dataset. The proposed unified framework can easily generate a model combined with preprocessing functions and lightweight artificial intelligence engine. In performance evaluation using handwritten image dataset and fall detection dataset measured with inertial sensor, the proposed unified framework showed building optimal artificial intelligence models with over 90% test accuracy.

경량 인공지능 하드웨어는 다양한 문제의 해결을 위해 멀티모달 센서 데이터를 입력받아 특징 선택, 추출, 차원축소, 정규화 과정을 수행한 후 인공지능 엔진으로 예측 결과를 도출한다. 다양한 애플리케이션에서 높은 성능을 달성하기 위해서는 이러한 경량 인공지능 하드웨어의 초 매개변수와 전체적인 전처리 시스템의 구성을 데이터에 맞춰 최적화할 필요가 있다. 본 논문에서는 경량 인공지능 하드웨어의 효율적인 제어 및 최적화를 위한 통합 프레임워크를 제안한다. 제안된 통합 프레임워크는 데이터 전처리 및 뉴로모픽 기반 경량 인공지능 엔진을 유연하게 재구성할 수 있으며, 최적의 모델을 생성할 수 있다. 기능검증을 위해 손글씨 이미지 데이터 세트와 관성 센서 데이터 기반의 낙상 검출 데이터 세트를 사용하였으며, 실험 결과 제안하는 통합 프레임워크가 각각의 데이터 세트에서 90% 이상의 정확도를 갖는 최적의 모델을 생성함을 확인하였다.

Keywords

References

  1. Gartner Research, "The Hyper-connected Enterprise: Anticipating the Next Wave of Business," Technical report, Feb. 2008.
  2. Gartner Research, "Gartner Identifies the Top 10 Strategic Technology Trends for 2019," Technical report, Oct. 2018.
  3. B. Kim, J. Lee, T. Hwang, and D. Kim "Design of Lightweight Artificial Intelligence System for Multimodal Signal Processing," J. of the Korea Institute of Electronic Communication Sciences, vol. 13, no. 5, 2018, pp. 1037-1042. https://doi.org/10.13067/JKIECS.2018.13.5.1037
  4. A. Ozdemir and B. Barshan "Detecting Falls with Wearable Sensors Using Machine Learning Techniques," J. of Sensors, vol. 14, no. 6, Apr. 2014, pp. 10691-10708. https://doi.org/10.3390/s140610691
  5. T. Baltrusaitis, C. Ahuja, and L. Morency, "Multimodal Machine Learning: A Survey and Taxonomy," arXiv preprint arXiv:1705.09406, 2017, pp. 1-20.
  6. J. Benediktsson and J. Sveinsson, "Feature extraction for neural network classifiers," Neurocomputation in Remote Sensing Data Analysis, Berlin, Heidelberg, Germany, 1997, pp. 97-104.
  7. J. Cho, Y. Jung, S. Lee, and Y. Jung, "VLSI Implementation of Restricted Coulomb Energy Neural Network with Improved Learning Scheme," MDPI Electronics, vol. 8, no. 563, 2019, pp. 68-73.
  8. H. Kim "A Study on The Real-Time Data Collection/Analysis/Processing Intelligent IoT," J. of the Korea Institute of Electronic Communication Sciences, vol. 14, no. 2, 2019, pp. 317-322. https://doi.org/10.13067/JKIECS.2019.14.2.317
  9. J. Lee, B. Kim, H. Park, D. Kim, and J. Kwon "Genetic Algorithm Calibration Method and PnP Platform for Multimodal Sensor," J. of the Korea Institute of Electronic Communication Sciences, vol. 14, no. 1, 2019, pp. 69-80. https://doi.org/10.13067/JKIECS.2019.14.1.69
  10. M. Buhmann, Radial Basis Functions: Theory and Implementations. Cambridge, United Kingdom, 2009.
  11. UCI Machine Learning Repository, "Optical Recognition of Handwritten Digits Data Set," July 1998.
  12. UCI Machine Learning Repository, "Simulated Falls and Daily Living Activities Data Set," June 2018.