A study of dynamic meta-learning for failure prediction in large-scale systems

Despite years of study on failure prediction, it remains an open problem, especially in large-scale systems composed of vast amount of components. In this paper, we present a dynamic meta-learning framework for failure prediction. It intends to not only provide reasonable prediction accuracy, but also be of practical use in realistic environments. Two key techniques are developed to address technical challenges of failure prediction. One is meta-learning to boost prediction accuracy by combining the benefits of multiple predictive techniques. The other is a dynamic approach to dynamically obtain failure patterns from a changing training set and to dynamically extract effective rules by actively monitoring prediction accuracy at runtime. We demonstrate the effectiveness and practical use of this framework by means of real system logs collected from the production Blue Gene/L systems at Argonne National Laboratory and San Diego Supercomputer Center. Our case studies indicate that the proposed mechanism can provide reasonable prediction accuracy by forecasting up to 82% of the failures, with a runtime overhead less than 1.0 min.     

A combination of hidden Markov model and fuzzy model for stock market forecasting

This paper presents a novel combination of the hidden Markov model (HMM) and the fuzzy models for forecasting stock market data. In a previous study we used an HMM to identify similar data patterns from the historical data and then used a weighted average to generate a ‘one-day-ahead’ forecast. This paper uses a similar approach to identify data patterns by using the HMM and then uses fuzzy logic to obtain a forecast value. The HMM's log-likelihood for each of the input data vectors is used to partition the dataspace. Each of the divided dataspaces is then used to generate a fuzzy rule. The fuzzy model developed from this approach is tested on stock market data drawn from different sectors. Experimental results clearly show an improved forecasting accuracy compared to other forecasting models such as, ARIMA, artificial neural network (ANN) and another HMM-based forecasting model.     

A deep spatio-temporal meta-learning model for urban traffic revitalization index prediction in the COVID-19 pandemic

The COVID-19 pandemic is a major global public health problem that has caused hardship to people’s normal production and life. Predicting the traffic revitalization index can provide references for city managers to formulate policies related to traffic and epidemic prevention. Previous methods have struggled to capture the complex and diverse dynamic spatio-temporal correlations during the COVID-19 pandemic. Therefore, we propose a deep spatio-temporal meta-learning model for the prediction of traffic revitalization index (DeepMeta-TRI) using external auxiliary information such as COVID-19 data. We conduct extensive experiments on a real-world dataset, and the results validate the predictive performance of DeepMeta-TRI and its effectiveness in addressing underfitting.     

Combining meta-learning and search techniques to select parameters for support vector machines

Support Vector Machines (SVMs) have achieved very good performance on different learning problems. However, the success of SVMs depends on the adequate choice of the values of a number of parameters (e.g., the kernel and regularization parameters). In the current work, we propose the combination of meta-learning and search algorithms to deal with the problem of SVM parameter selection. In this combination, given a new problem to be solved, meta-learning is employed to recommend SVM parameter values based on parameter configurations that have been successfully adopted in previous similar problems. The parameter values returned by meta-learning are then used as initial search points by a search technique, which will further explore the parameter space. In this proposal, we envisioned that the initial solutions provided by meta-learning are located in good regions of the search space (i.e. they are closer to optimum solutions). Hence, the search algorithm would need to evaluate a lower number of candidate solutions when looking for an adequate solution. In this work, we investigate the combination of meta-learning with two search algorithms: Particle Swarm Optimization and Tabu Search. The implemented hybrid algorithms were used to select the values of two SVM parameters in the regression domain. These combinations were compared with the use of the search algorithms without meta-learning. The experimental results on a set of 40 regression problems showed that, on average, the proposed hybrid methods obtained lower error rates when compared to their components applied in isolation.     

A metric-based meta-learning approach combined attention mechanism and ensemble learning for few-shot learning

Meta-learning is one of the latest research directions in machine learning, which is considered to be one of the most probably ways to realize strong artificial intelligence. Meta-learning focuses on seeking solutions for machines to learn like human beings do - to recognize things through only few sample data and quickly adapt to new tasks. Challenges occur in how to train an efficient machine model with limited labeled data, since the model is easily over-fitted. In this paper, we address this obvious but important problem and propose a metric-based meta-learning model, which combines attention mechanisms and ensemble learning method. In our model, we first design a dual path attention module which considers both channel attention and spatial attention module, and the attention modules have been stacked to conduct a meta-learner for few shot meta-learning. Then, we apply an ensemble method called snap-shot ensemble to the attention-based meta-learner in order to generate more models in a single episode. Features abstracted from the models are put into the metric-based architecture to compute a prototype for each class. Our proposed method intensifies the feature extracting ability of backbone network in meta-learner and reduces over-fitting through ensemble learning and metric learning method. Experimental results toward several meta-learning datasets show that our approach is effective.     

The forecasting ability of Internet-based virtual futures market

Internet-based virtual futures markets (VFMs) have been used in predicting election results and movie ticket sales. We construct an Internet-based VFM to predict an underlying stock price. While the virtual futures market has received much attention, questions remain as to the ideal number of participants. Results of Granger causality tests and analysis of directional accuracy show that a VFM with only a small number of participants (75) is able to generate informative futures prices useful in the prediction of the underlying stock price. Moreover, the participants were not professional investors but merely undergraduate finance students with only a cursory introduction to futures trading. Our results provide additional evidence supporting the use of VFMs in forecasting and show that VFMs are powerful forecasting tools.     

A FCM-based deterministic forecasting model for fuzzy time series

The study of fuzzy time series has increasingly attracted much attention due to its salient capabilities of tackling uncertainty and vagueness inherent in the data collected. A variety of forecasting models including high-order models have been devoted to improving forecasting accuracy. However, the high-order forecasting approach is accompanied by the crucial problem of determining an appropriate order number. Consequently, such a deficiency was recently solved by Li and Cheng [S.-T. Li, Y.-C. Cheng, Deterministic Fuzzy time series model for forecasting enrollments, Computers and Mathematics with Applications 53 (2007) 1904–1920] using a deterministic forecasting method. In this paper, we propose a novel forecasting model to enhance forecasting functionality and allow processing of two-factor forecasting problems. In addition, this model applies fuzzy c-means (FCM) clustering to deal with interval partitioning, which takes the nature of data points into account and produces unequal-sized intervals. Furthermore, in order to cope with the randomness of initially assigned membership degrees of FCM clustering, Monte Carlo simulations are used to justify the reliability of the proposed model. The superior accuracy of the proposed model is demonstrated by experiments comparing it to other existing models using real-world empirical data.     

A hybrid fuzzy regression-fuzzy cognitive map algorithm for forecasting and optimization of housing market fluctuations

This paper presents a hybrid algorithm based on fuzzy linear regression (FLR) and fuzzy cognitive map (FCM) to deal with the problem of forecasting and optimization of housing market fluctuations. Due to the uncertainty and severe noise associated with the housing market, the application of crisp data for forecasting and optimization purposes is insufficient. Hence, in order to enable the decision-makers to make decisions with respect to imprecise/fuzzy data, FLR is used in the proposed hybrid algorithm. The best-fitted FLR model is then selected with respect to two indicators including Index of Confidence (IC) and Mean Absolute Percentage Error (MAPE). To achieve this objective, analysis of variance (ANOVA) for a randomized complete block design (RCBD) is employed. The primary objective of this study is to utilize imprecise/fuzzy data in order to improve the analysis of housing price fluctuations, in accordance with the factors obtained through the best-fitted FLR model. The secondary objective of this study is the exhibition of the resulted values in a schematic way via FCM. Hybridization of FLR and FCM provides a decision support system (DSS) for utilization of historical data to predict housing market fluctuation in the future and identify the influence of the other parameters. The proposed hybrid FLR-FCM algorithm enables the decision-makers to utilize imprecise and ambiguous data and represent the resulted values of the model more clearly. This is the first study that utilizes a hybrid intelligent approach for housing price and market forecasting and optimization.     

Deterministic fuzzy time series model for forecasting enrollments

The fuzzy time series has recently received increasing attention because of its capability of dealing with vague and incomplete data. There have been a variety of models developed to either improve forecasting accuracy or reduce computation overhead. However, the issues of controlling uncertainty in forecasting, effectively partitioning intervals, and consistently achieving forecasting accuracy with different interval lengths have been rarely investigated. This paper proposes a novel deterministic forecasting model to manage these crucial issues. In addition, an important parameter, the maximum length of subsequence in a fuzzy time series resulting in a certain state, is deterministically quantified. Experimental results using the University of Alabama’s enrollment data demonstrate that the proposed forecasting model outperforms the existing models in terms of accuracy, robustness, and reliability. Moreover, the forecasting model adheres to the consistency principle that a shorter interval length leads to more accurate results.     

A two-stage dynamic sales forecasting model for the fashion retail

The difficulty with fashion retail forecasting is due to a number of factors such as the season, region and fashion effect and causes a nonlinear change in the original sales rules. To improve the accuracy of fashion retail forecasting, a two-stage dynamic forecasting model is proposed, which is combined with both long-term and short-term predictions. The model introduces the improved adjustment methods, the main adjustment model and error forecasting model in the adjustment system collaborated with each other. The real-time data are demonstrated by applying the model in wireless mobile environment. The experiment shows that the model provides good results for fashion retail forecasting.     

基于隐马尔可夫模型的股票价格预测组合模型

提出了一种用于股票价格预测的人工神经网络(ANN),隐马尔可夫模型(HMM)和粒子群优化算法(PSO)的组合模型-APHMM模型.在APHMM模型中,ANN算法将股票的每日开盘价、最高价、最低价与收盘价转换为相互独立的量并作为HMM的输入.然后,利用PSO算法对HMM的参数初始值进行优化,并用Baum-Welch算法进行参数训练.经过训练后的HMM在历史数据中找出一组与今天股票的上述4个指标模式最相似数据,加权平均计算每个数据与它后一天的收盘价格差,则今天的股票收盘价加上这个加权平均价格差便为预测的股票收盘价.实验结果表明,APHMM模型具有良好的预测性能.     

A hybrid ANFIS model based on empirical mode decomposition for stock time series forecasting

Time series forecasting is an important and widely popular topic in the research of system modeling, and stock index forecasting is an important issue in time series forecasting. Accurate stock price forecasting is a challenging task in predicting financial time series. Time series methods have been applied successfully to forecasting models in many domains, including the stock market. Unfortunately, there are 3 major drawbacks of using time series methods for the stock market: (1) some models can not be applied to datasets that do not follow statistical assumptions; (2) most time series models that use stock data with a significant amount of noise involutedly (caused by changes in market conditions and environments) have worse forecasting performance; and (3) the rules that are mined from artificial neural networks (ANNs) are not easily understandable.To address these problems and improve the forecasting performance of time series models, this paper proposes a hybrid time series adaptive network-based fuzzy inference system (ANFIS) model that is centered around empirical mode decomposition (EMD) to forecast stock prices in the Taiwan Stock Exchange Capitalization Weighted Stock Index (TAIEX) and Hang Seng Stock Index (HSI). To measure its forecasting performance, the proposed model is compared with Chen's model, Yu's model, the autoregressive (AR) model, the ANFIS model, and the support vector regression (SVR) model. The results show that our model is superior to the other models, based on root mean squared error (RMSE) values.     

A hybrid fuzzy and neural approach for DRAM price forecasting

The trend in the price of dynamic random access memory (DRAM) is a very important prosperity index in the semiconductor industry. To further enhance the performance of DRAM price forecasting, a hybrid fuzzy and neural approach is proposed in this study. In the proposed approach, multiple experts construct their own fuzzy multiple linear regression models from various viewpoints to forecast the price of a DRAM product. Each fuzzy multiple linear regression model can be converted into two equivalent nonlinear programming problems to be solved. To aggregate these fuzzy price forecasts, a two-step aggregation mechanism is applied. At the first step, fuzzy intersection is applied to aggregate the fuzzy price forecasts into a polygon-shaped fuzzy number, in order to improve the precision. After that, a back propagation network is constructed to defuzzify the polygon-shaped fuzzy number and to generate a representative/crisp value, so as to enhance the accuracy. A real example is used to evaluate the effectiveness of the proposed methodology. According to experimental results, the proposed methodology improved both the precision and accuracy of DRAM price forecasting by 66% and 43%, respectively.     

A dynamic trading rule based on filtered flag pattern recognition for stock market price forecasting

In this paper we propose and validate a trading rule based on flag pattern recognition, incorporating important innovations with respect to the previous research. Firstly, we propose a dynamic window scheme that allows the stop loss and take profit to be updated on a quarterly basis. In addition, since the flag pattern is a trend-following pattern, we have added the EMA indicator to filter trades. This technical analysis indicator is calculated both for 15-min and 1-day timeframes, which enables short and medium terms to be considered simultaneously. We also filter the flags according to the price range on which they are developed and have limited the maximum loss of each trade to 100 points. The proposed methodology was applied to 91,309 intraday observations of the DJIA index, considerably improving the results obtained in the previous proposals and those obtained by the buy & hold strategy, both for profitability and risk, and also after taking into account the transaction costs. These results seem to challenge market efficiency in line with other similar studies, in the specific analysis carried out on the DJIA index and is also limited to the setup considered.     

A collaborative and artificial intelligence approach for semiconductor cost forecasting

Forecasting the unit cost of a semiconductor product is an important task to the manufacturer. However, it is not easy to deal with the uncertainty in the unit cost. In order to effectively forecast the semiconductor unit cost, a collaborative and artificial intelligence approach is proposed in this study. In the proposed methodology, a group of domain experts is formed. These domain experts are asked to configure their own fuzzy neural networks to forecast the semiconductor unit cost based on their viewpoints. A collaboration mechanism is therefore established. To facilitate the collaboration process and to derive a single representative value from these forecasts, a radial basis function (RBF) network is used. The effectiveness of the proposed methodology is shown with a case study.     

Variable weights combined model based on multi-objective optimization for short-term wind speed forecasting

Accurate and steady wind speed prediction is essential for the efficient management of wind power factories and energy systems. However, it is difficult to obtain satisfactory forecasting performance because of the characteristics of random nonlinear fluctuations inherent in wind speed variation. Considering the drawbacks of statistical models in forecasting nonlinear time series and the problem of artificial intelligence models easily falling into a local optimum, in this study, we successfully integrate the variable weighted combination theory into a new combined forecasting model that simultaneously consists of three disparate hybrid models based on the decomposition technology. Moreover, the extreme learning machine optimized by the multi-objective grasshopper optimization algorithm is adopted to integrate all the forecasting results derived from each hybrid model to further enhance the forecasting accuracy. In this study, we consider a case study that employs several authentic wind speed data aggregates of Shandong wind farms for an evaluation of the forecasting performance of the proposed combined model. The experimental results reveal that this proposed model surpasses the contrasted benchmark models and is satisfactory for intellective grid programs.     

A variable spread fuzzy linear regression model with higher explanatory power and forecasting accuracy

Fuzzy regression models have been applied to operational research (OR) applications such as forecasting. Some of previous studies on fuzzy regression analysis obtain crisp regression coefficients for eliminating the problem of increasing spreads for the estimated fuzzy responses as the magnitude of the independent variable increases; however, they still cannot cope with the situation of decreasing or variable spreads. This paper proposes a three-phase method to construct the fuzzy regression model with variable spreads to resolve this problem. In the first phase, on the basis of the extension principle, the membership functions of the least-squares estimates of regression coefficients are constructed to conserve completely the fuzziness of observations. In the second phase, then they are defuzzified by the center of gravity method to obtain crisp regression coefficients. In the third phase, the error terms of the proposed model are determined by setting each estimated spread equals its corresponding observed spread. Furthermore, the Mamdani fuzzy inference system is adopted for improving the accuracy of its forecasts. Compared to the previous studies, the results from five examples and an application example of Japanese house prices show that the proposed fuzzy linear regression model has higher explanatory power and forecasting performance.     

A GA-weighted ANFIS model based on multiple stock market volatility causality for TAIEX forecasting

Stock market forecasting is important and interesting, because the successful prediction of stock prices may promise attractive benefits. The economy of Taiwan relies on international trade deeply, and the fluctuations of international stock markets will impact Taiwan stock market. For this reason, it is a practical way to use the fluctuations of other stock markets as forecasting factors for forecasting the Taiwan stock market. In this paper, the proposed model uses the fluctuations of other national stock markets as forecasting factors and employs a genetic algorithm (GA) to refine the weights of rules joining in an ANFIS model to forecast the Taiwan stock index. To evaluate the forecasting performances, the proposed model is compared with four different models: Chen's model, Yu's model, Huarng's model, and the ANFIS model. The results indicate that the proposed model is superior to the listing methods in terms of the root mean squared error (RMSE).     

Evaluation of forecasting methods for intermittent parts demand in the field of aviation: a predictive model

Owing to the sporadic nature of demand for aircraft maintenance repair parts, airline operators perceive difficulties in forecasting and are still looking for superior forecasting methods. This paper deals with techniques applicable to predicting spare parts demand for airline fleets. The experimental results of 13 forecasting methods, including those used by aviation companies, are examined and clarified through statistical analysis. The general linear model approach is used to explain the variation attributable to different experimental factors and their interactions. Actual historical data for hard-time and condition-monitoring components from an airlines operator are used, in order to compare different forecasting methods when facing intermittent demand. The results confirm the continued superiority of the weighted moving average, Holt and Croston method for intermittent demand, whereas most commonly used methods by airlines are found to be questionable, consistently producing poor forecasting performance. We have, however, devised a new approach to forecasting evaluation, a predictive error-forecasting model which compares and evaluates forecasting methods based on their factor levels when faced with intermittent demand. A simple example is presented to illustrate the performance of the mathematical model. It is suggested that these findings may be applicable to other industrial sectors, which have similar demand patterns to those of airlines.Scope and purposeDemand forecasting is one of the most crucial issues of inventory management. Forecasts, which form the basis for the planning of inventory levels, are probably the biggest challenge in the repair and overhaul industry, as the one common problem facing airlines throughout the world is the need to know the short-term part demand forecast with the highest possible degree of accuracy. The high cost of modern aircraft and the expense of such repairable spares as aircraft engines and avionics constitute a large part of the total investment of many airline operators. These parts, though low in demand, are critical to operations and their unavailability can lead to excessive down time costs. Most airline materials managers deal with intermittent demand, which tends to be random and has a large proportion of zero values. In an effort to achieve this, the study has presented a model that could be of great benefit to airline operators and other maintenance service organisations. It will enable them to select in advance the appropriate forecasting method that better meets their cyclical demand for parts. This approach is consistent with the purpose of this study, which aims to compare different forecasting methods when faced with intermittent demand.