SOM-based partial labeling of imbalanced data stream

Filtering-Based Instance Selection Method for Overlapping Problem in Imbalanced Datasets

The overlapping problem occurs when a region of the dimensional data space is shared in a similar proportion by different classes. It has an impact on a classifier’s performance due to the difficulty in correctly separating the classes. Further, an imbalanced dataset consists of a situation in which one class has more instances than another, and this is another aspect that impacts a classifier’s performance. In general, these two problems are treated separately. On the other hand, Prototype Selection (PS) approaches are employed as strategies for selecting appropriate instances from a dataset by filtering redundant and noise data, which can cause misclassification performance. In this paper, we introduce Filtering-based Instance Selection (FIS), using as a base the Self-Organizing Maps Neural Network (SOM) and information entropy. In this sense, SOM is trained with a dataset, and, then, the instances of the training set are mapped to the nearest prototype (SOM neurons). An analysis with entropy is conducted in each prototype region. From a threshold, we propose three decision methods: filtering the majority class (H-FIS (High Filter IS)), the minority class (L-FIS (Low Filter IS)), and both classes (B-FIS). The experiments using artificial and real dataset showed that the methods proposed in combination with 1NN improved the accuracy, F-Score, and G-mean values when compared with the 1NN classifier without the filter methods. The FIS approach is also compatible with the approaches mentioned in the relevant literature.

Dynamic Long Short-Term Memory Neural-Network- Based Indirect Remaining-Useful-Life Prognosis for Satellite Lithium-Ion Battery

On-line remaining-useful-life (RUL) prognosis is still a problem for satellite Lithium-ion (Li-ion) batteries. Meanwhile, capacity, widely used as a health indicator of a battery (HI), is inconvenient or even impossible to measure. Aiming at practical and precise prediction of the RUL of satellite Li-ion batteries, a dynamic long short-term memory (DLSTM) neural-network-based indirect RUL prognosis is proposed in this paper. Firstly, an indirect HI based on the Spearman correlation analysis method is extracted from the battery discharge voltages, and the relationship between the indirect HI indices and battery capacity is established using a polynomial fitting method. Then, by integrating the Adam method, L2 regularization method, and incremental learning, a DLSTM method is proposed and applied for Li-ion battery RUL prognosis. Finally, verification of the results on NASA #5 battery data sets demonstrates that the proposed method has better dynamic performance and higher accuracy than the three other popular methods.