Predicting the Learning Performance of Artificial Intelligent Systems Using Non-Homogeneous Poisson Process Models


Artificial intelligent systems can learn to adapt to environmental changes to find a better solution. Improving system performance has been a great interest of study with new objective functions and parameters being constantly applied. However, approaches for performance evaluation are often by real observation without offering a prediction capability to support decision making at run time. Prediction helps foresee the future, so that a good run can continue while a poor one can be replaced. It also assists in evaluating the efficacy of different algorithms, especially when their learning capabilities vary over time. In this paper, a statistical approach is proposed to predict run time system learning performance. We apply non-homogeneous Poisson process (NHPP) models that account for time-variant learning effectiveness to determine the future solution more precisely. Our paradigm predicts the probability for the next improvement by time. Together with the estimated extent of that improvement, the expected performance can be measured. Learning algorithms can be compared based on their performance metrics.

  • Abstract
  • 1. Introduction
  • 2. Methodology
  • 3. A Case Study
  • 4. Conclusion
  • References

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In