Training Radial Basis Function Neural Networks for Classification via Class-Specific Clustering

Predicting demand for long-term care using Japanese healthcare insurance claims data

BACKGROUND

Driven by the rapid aging of the population, Japan introduced public long-term care insurance to reinforce healthcare services for the elderly in 2000. Precisely predicting future demand for long-term care services helps authorities to plan and manage their healthcare resources and citizens to prevent their health status deterioration.

METHODS

This paper presents our novel study for developing an effective model to predict individual-level future long-term care demand using previous healthcare insurance claims data. We designed two discriminative models and subsequently trained and validated the models using three learning algorithms with medical and long-term care insurance claims and enrollment records, which were provided by 170 regional public insurers in Gifu, Japan.

RESULTS

The prediction model based on multiclass classification and gradient-boosting decision tree achieved practically high accuracy (weighted average of Precision, 0.872; Recall, 0.878; and F-measure, 0.873) for up to 12 months after the previous claims. The top important feature variables were indicators of current health status (e.g., current eligibility levels and age), risk factors to worsen future healthcare status (e.g., dementia), and preventive care services for improving future healthcare status (e.g., training and rehabilitation).

CONCLUSIONS

The intensive validation tests have indicated that the developed prediction method holds high robustness, even though it yields relatively lower accuracy for specific patient groups with health conditions that are hard to distinguish.

Intelligent modeling and monitoring of micro-droplet profiles in 3D printing

The inkjet 3D printing has been one of the most studied and applied additive manufacturing (AM) processes in electronic industry. In this AM process, the forming quality is greatly influenced by the micro-droplet deposition and substrate temperature. While most studies focus on the formation mechanism of droplets, there are few studies on the quantitative evaluation of the droplet surface profile and its qualitative correlation with temperature changes. In this study, the characteristics of droplet profile in three-dimensional inkjet printing were studied from two aspects, the modeling of droplet shape and the estimation of droplet temperature. For this purpose, different types of radial basis function networks (RBFN) are applied. The validity of the regularized RBFN model is developed and verified by experiments. The results show that the droplet shape can be accurately modeled and the drying temperature can be accurately estimated given the model.

Radial Basis Function Neural Network with Localized Stochastic-Sensitive Autoencoder for Home-Based Activity Recognition

Over the past few years, the Internet of Things (IoT) has been greatly developed with one instance being smart home devices gradually entering into people's lives. To maximize the impact of such deployments, home-based activity recognition is required to initially recognize behaviors within smart home environments and to use this information to provide better health and social care services. Activity recognition has the ability to recognize people's activities from the information about their interaction with the environment collected by sensors embedded within the home. In this paper, binary data collected by anonymous binary sensors such as pressure sensors, contact sensors, passive infrared sensors etc. are used to recognize activities. A radial basis function neural network (RBFNN) with localized stochastic-sensitive autoencoder (LiSSA) method is proposed for the purposes of home-based activity recognition. An autoencoder (AE) is introduced to extract useful features from the binary sensor data by converting binary inputs into continuous inputs to extract increased levels of hidden information. The generalization capability of the proposed method is enhanced by minimizing both the training error and the stochastic sensitivity measure in an attempt to improve the ability of the classifier to tolerate uncertainties in the sensor data. Four binary home-based activity recognition datasets including OrdonezA, OrdonezB, Ulster, and activities of daily living data from van Kasteren (vanKasterenADL) are used to evaluate the effectiveness of the proposed method. Compared with well-known benchmarking approaches including support vector machine (SVM), multilayer perceptron neural network (MLPNN), random forest and an RBFNN-based method, the proposed method yielded the best performance with 98.35%, 86.26%, 96.31%, 92.31% accuracy on four datasets, respectively.

Radial Basis Functions With Adaptive Input and Composite Trend Representation for Portfolio Selection

We propose a set of novel radial basis functions with adaptive input and composite trend representation (AICTR) for portfolio selection (PS). Trend representation of asset price is one of the main information to be exploited in PS. However, most state-of-the-art trend representation-based systems exploit only one kind of trend information and lack effective mechanisms to construct a composite trend representation. The proposed system exploits a set of RBFs with multiple trend representations, which improves the effectiveness and robustness in price prediction. Moreover, the input of the RBFs automatically switches to the best trend representation according to the recent investing performance of different price predictions. We also propose a novel objective to combine these RBFs and select the portfolio. Extensive experiments on six benchmark data sets (including a new challenging data set that we propose) from different real-world stock markets indicate that the proposed RBFs effectively combine different trend representations and AICTR achieves state-of-the-art investing performance and risk control. Besides, AICTR withstands the reasonable transaction costs and runs fast; hence, it is applicable to real-world financial environments.

Image-Specific Classification With Local and Global Discriminations

Most image classification methods try to learn classifiers for each class using training images alone. Due to the interclass and intraclass variations, it would be more effective to take the testing images into consideration for classifier learning. In this brief, we propose a novel image-specific classification method by combing the local and global discriminations of training images. We adaptively train classifier for each testing image instead of generating classifiers for each class with training images alone. For each testing image, we first select its ${k}$ nearest neighbors in the training set with the corresponding labels for local classifier training. This helps to model the distinctive characters of each testing image. Besides, we also use all the training images for global discrimination modeling. The local and global discriminations are combined for final classification. In this way, we could not only model the specific character of each testing image but also avoid the local optimum by jointly considering all the training images. To evaluate the usefulness of the proposed image-specific classification with local and global discrimination (ISC-LG) method, we conduct image classification experiments on several public image data sets. The superior performances over other baseline methods prove the effectiveness of the proposed ISC-LG method.

A Peak Price Tracking-Based Learning System for Portfolio Selection

We propose a novel linear learning system based on the peak price tracking (PPT) strategy for portfolio selection (PS). Recently, the topic of tracking control attracts intensive attention and some novel models are proposed based on backstepping methods, such that the system output tracks a desired trajectory. The proposed system has a similar evolution with a transform function that aggressively tracks the increasing power of different assets. As a result, the better performing assets will receive more investment. The proposed PPT objective can be formulated as a fast backpropagation algorithm, which is suitable for large-scale and time-limited applications, such as high-frequency trading. Extensive experiments on several benchmark data sets from diverse real financial markets show that PPT outperforms other state-of-the-art systems in computational time, cumulative wealth, and risk-adjusted metrics. It suggests that PPT is effective and even more robust than some defensive systems in PS.