No abstract

Forecasting the future states of a complex system is a complicated challenge that is encountered in many industrial applications covered in the community of prognostics and health management. Practically, states can be either continuous or discrete: Continuous states generally represent the value of a signal while discrete states generally depict functioning modes reflecting the current degradation. For each case, specific techniques exist. In this paper, we propose an approach based on case-based reasoning that jointly estimates the future values of the continuous signal and the future discrete modes. The main characteristics of the proposed approach are the following: 1) It relies on the K-nearest neighbor algorithm based on belief function theory; 2) belief functions allow the user to represent his/her partial knowledge concerning the possible states in the training data set, particularly concerning transitions between functioning modes which are imprecisely known; and 3) two distinct strategies are proposed for state prediction, and the fusion of both strategies is also considered. Two real data sets were used in order to assess the performance in estimating future breakdown of a real system.

In many real-world applications, ranging from predictive maintenance to personalized medicine, early classification of time series data is of paramount importance for supporting decision makers. In this article, we address this challenging task with a novel approach based on reinforcement learning. We introduce an early classifier agent, an end-to-end reinforcement learning agent (deep Q-network, DQN) [1] able to perform early classification in an efficient way. We formulate the early classification problem in a reinforcement learning framework: we introduce a suitable set of states and actions but we also define a specific reward function which aims at finding a compromise between earliness and classification accuracy. While most of the existing solutions do not explicitly take time into account in the final decision, this solution allows the user to set this trade-off in a more flexible way. In particular, we show experimentally on datasets from the UCR time series archive [2] that this agent is able to continually adapt its behavior without human intervention and progressively learn to compromise between accurate and fast predictions.