Predictive and decision models in finance are typically validated under assumptions of distributional stability over the evaluation window. In deployment, those assumptions fail: the data-generating process undergoes structural change—breaks, regime transitions, and drift—that can invalidate conditional relationships, degrade calibration, and amplify tail risk precisely when decisions are most consequential. Despite a large literature, results remain hard to reconcile across econometrics, statistical monitoring, and machine learning due to divergent terminology and incompatible evaluation protocols. This survey addresses this fragmentation through a taxonomy-based methodological synthesis of financial non-stationarity, linking drift signatures to design trade-offs in representation, detection, adaptation, and evaluation. We contribute: (1) a unified taxonomy of drift and regime change; (2) a structured, literature-grounded synthesis of drift-aware representation, change detection, and continuous adaptation methods; and (3) scenario-based evaluation guidance to support future comparative studies under non-stationarity in financial time series. We align terms such as structural breaks, regimes, concept drift, and dataset shift, and organize them through a taxonomy that combines temporal, statistical, spatial, ontological, and causal dimensions to describe real drift scenarios consistently. Using this lens, we review drift-aware representations, change detection methods, and continuous adaptation strategies—from classical sequential monitoring and segmentation to Bayesian, multivariate, and embedding-based out-of-distribution approaches. We then consolidate evaluation guidance spanning detection delay, false-alarm control, computational cost, and finance-specific utility. Finally, we highlight emerging directions (foundation models, multimodal context, parameter-efficient adaptation) and open challenges in benchmark design and reliable online calibration.

This paper introduces the Trend-Enhanced CNN Adversarial Autoencoder (TE-CNN-AAE), a novel approach for generating low-dimensional representations of intraday stock market activity from 5-minute interval quotes. The proposed model extends traditional autoencoders by incorporating an adversarial component that enhances the quality of embeddings by reconstructing input sequences while simultaneously estimating market trends. Using five years of intraday data from Dow Jones Industrial Average (DJIA) assets, we demonstrate that TE-CNN-AAE outperforms baseline methods in capturing intraday patterns and predicting daily price movements. Qualitative evaluations via UMAP visualizations show improved class separability in the latent space, further confirmed by the Silhouette Score and Davies-Bouldin Index, while quantitative analysis using an LSTM classifier validates the superior predictive utility of the embeddings. Ablation tests confirm the importance of the adversarial discriminator in generating robust representations. Our results suggest that TE-CNN-AAE effectively captures the complex dynamics of financial time series and holds the potential for improving decision-support systems in trading.

Just-in-time Software Defect Prediction (JIT-SDP) aims to detect potential defects early, helping to prevent risky code from entering the repository during development. This study evaluates JIT-SDP using pre-trained language models in different architectures and settings. It compares open-source fine-tuned models such as CodeT5+ and UniXCoder with closed LLMs such as GPT and Gemini. This is the first known study to compare trainable open and prompt-based closed decoder-only models for JIT-SDP. The main results show that fine-tuned open models outperform closed models in zero-shot and few-shot scenarios without advanced prompt engineering techniques, and in cross-project tasks, CodeT5+ and UniXCoder surpass previous state-of-the-art results. The findings underscore the value of model architecture, fine-tuning, and expert features for effective defect prediction. Finally, we introduce CodeFlowLM – to our knowledge, the first framework for continual JIT-SDP using pre-trained language models.

In the evolving landscape of software engineering, accurate identification of defect-inducing commits is critical to improving software quality and reducing development costs. This paper revisits the widely adopted SZZ algorithm, which is utilized for labeling commits as clean or defect-inducing, to address one of its main limitations, i.e., its reliance on outdated corrective commits identification strategies. We propose an innovative approach that integrates the semantic understanding capability of the GPT OpenAI model into the SZZ flow to better interpret commit messages. Our experiments reveal, for some projects, a large number of commits incorrectly interpreted as defect-fixing, consequently, leading to the misclassification of commits as defect-inducing. As an example, for the Postgresql dataset, the number of defect-inducing commits was reduced in 21% when compared to the original SZZ. Furthermore, results of our experiments strongly suggest that, as a result of the proposed SZZ labeling process, the JIT-SDP problem has been shown to be more challenging than originally reported by previous works.

While the University of Birmingham exercises care and attention in making items available there are rare occasions when an item has been uploaded in error or has been deemed to be commercially or otherwise sensitive.If you believe that this is the case for this document, please contact [email protected] providing details and we will remove

Deep learning models often require large datasets to perform well from scratch. Transfer learning methods solve this issue by using a pre-trained source network to improve a target network training. Recent approaches involve using feature maps from the source network to guide the target network training. The latest transfer learning methods use meta-networks to enhance the knowledge transfer process. These meta-networks bridge the source and target networks, deciding which pairs of feature map layers and channels should be matched for optimal knowledge transfer. This paper improves this approach by using pixel-level information, in addition to layers and channels, for better knowledge transfer. Our experiments on multiple datasets show that the proposed approach outperforms previous baselines in scenarios with limited labels per class. The source code is available at https://github.com/lucasdelimanogueira/L2T-www.

Context: Process Mining (PM) aims to discover processes and their characteristics from event logs recorded by information systems. There are dozens of general-purpose tools. The Brazilian judiciary wants to make the technology available to magistrates with little or no knowledge of the field of PM. Problem: The usability and comprehensibility of the available tools prevent their adoption by laypeople. In fact, these are two of the eleven challenges proposed by the IEEE Task Force on PM that are still little explored in the context of non-specialists. Methodology: Applied qualitative research, using User Centered Design (UCD) principles to guide the construction of the JuMP tool, conceived considering anthropological and sociological aspects of the Brazilian judiciary. For a year and six months, a team of PM specialists worked with representatives of the judiciary sector to produce, evaluate and evolve the new product. An experiment was performed to evaluate the usability and understandability of JuMP. Results: The study demonstrated that the use of a PM tool oriented to the application domain is fundamental for domain experts with little or no knowledge in PM to be able to make good use of it. Contribution: Demonstration that, unlike the prevailing practice in the area of PM, which prioritizes the provision of general purpose tools, the design of tools oriented to the application domain are prerequisites to improve usability and comprehensibility attributes. Finally, the work proposes a generic and adaptable methodology for developing specific-purpose PM tools and the experience of its implications to the design of the JuMP tool.

Abstract Just-in-Time Software Defect Prediction (JIT-SDP) operates in an online scenario where additional training data is received over time. Existing online JIT-SDP studies used online Oza ensemble learning methods with Hoeffding Trees as base learners to learn and update JIT-SDP models over time in this scenario. However, it is unknown how these approaches compare against offline learning approaches adapted to operate in online scenarios, and how the use of any other online or offline base learners would affect online JIT-SDP in terms of predictive performance and computational cost. We therefore propose a new approach called Batch Oversampling Rate Boosting (BORB) that is able to use offline base learners in an online JIT-SDP scenario. Based on 10 open source projects, we provide a comprehensive evaluation of BORB with 5 different base learners and the existing online approach Oversampling Rate Boosting with 4 different base learners, both in within-project and cross-project online JIT-SDP scenarios. The results show that offline learning can lead to better predictive performance than the top performing online learning approaches considered in our study, at a higher computational cost. Cross-project data was helpful to improve predictive performance both for offline and online learning, but especially for online learning.

Real-world applications have been dealing with large amounts of data that arrive over time and generally present changes in their underlying joint probability distribution, i.e., concept drift. Concept drift can be subdivided into two types: virtual drift, which affects the unconditional probability distribution p(x), and real drift, which affects the conditional probability distribution p(y|x). Existing works focuses on real drift. However, strategies to cope with real drift may not be the best suited for dealing with virtual drift, since the real class boundaries remain unchanged. We provide the first in depth analysis of the differences between the impact of virtual and real drifts on classifiers' suitability. We propose an approach to handle both drifts called On-line Gaussian Mixture Model With Noise Filter For Handling Virtual and Real Concept Drifts (OGMMF-VRD). Experiments with seven synthetics and seven real-world datasets show that OGMMF-VRD outperforms other approaches with separate mechanisms to deal with virtual and real drifts. It also has more stable rankings and smaller drops in performance during drifting periods than existing ensemble approaches, thus being more reliable for adoption in practice.

The robustness and resilience of the deep learning models offer consistent and competitive results in real-world applications. Despite its adaptability, the training and adjustment of the hyperparameters still demand knowledge and time from the designer. This paper proposes a simple and effective approach based on the Greedy Randomized Adaptive Search Procedure (GRASP) algorithm that we adapt to optimize deep neural networks models. We evaluated the performance of the proposed approach using the models Deep Feedforward Neural Network (DFNN) and TabNet, considering the Tabu Search algorithm as a baseline in five tabular datasets. Both optimization algorithms showed high performance regarding the (i) quality of the best solution, (ii) convergence, and (iii) local search. However, the adapted GRASP approach showed better results, optimizing the deep models in all datasets with statistical significance.

In writer-independent verification systems, a single model is trained for all users of the system using dissimilarity vectors obtained through a dichotomy transformation that converts a multi-class problem into a 2-class problem comprising: (i) the intra-class dissimilarity vectors computed from samples of the same user, (ii) the inter-class dissimilarity vectors computed from samples of different users. When mapping handwritten signature representations, it is desired to obtain well-separated dense clusters of signature representations for each user, in such a way that transformed intra-class dissimilarity vectors tend to be separated from the inter-class dissimilarity vectors. Moreover, since skilled forgeries resemble reference signatures, it is also desired to obtain skilled forgery dissimilarity vectors that are further away from the region of the intra-class dissimilarity vectors. In this work, it is hypothesized that an improved dissimilarity space can be achieved through a multi-task framework for learning handwritten signature feature representations based on deep contrastive learning. The proposed framework is composed of two objective-specific tasks; it does not use skilled forgeries for training. The first task aims to map signature examples of a given user in a dense cluster, while linearly separating the signature representations of different users. The second task aims to adjust forgery representations by adopting a contrastive loss with the ability to perform hard negative mining. Hard negatives are similar examples but from different classes that can be seen as artificially generated skilled forgeries for training. In a writer-independent verification approach, the model obtained with the proposed framework is evaluated in terms of the equal error rate on GPDS-300, CEDAR and MCYT-75 datasets. Experiments demonstrated a statistically significant improvement in signature verification compared to the state-of-the-art SigNet feature extraction method.

In this article, we argue that the unsatisfactory out-of-distribution (OOD) detection performance of neural networks is mainly due to the SoftMax loss anisotropy and propensity to produce low entropy probability distributions in disagreement with the principle of maximum entropy. On the one hand, current OOD detection approaches usually do not directly fix the SoftMax loss drawbacks, but rather build techniques to circumvent it. Unfortunately, those methods usually produce undesired side effects (e.g., classification accuracy drop, additional hyperparameters, slower inferences, and collecting extra data). On the other hand, we propose replacing SoftMax loss with a novel loss function that does not suffer from the mentioned weaknesses. The proposed IsoMax loss is isotropic (exclusively distance-based) and provides high entropy posterior probability distributions. Replacing the SoftMax loss by IsoMax loss requires no model or training changes. Additionally, the models trained with IsoMax loss produce as fast and energy-efficient inferences as those trained using SoftMax loss. Moreover, no classification accuracy drop is observed. The proposed method does not rely on outlier/background data, hyperparameter tuning, temperature calibration, feature extraction, metric learning, adversarial training, ensemble procedures, or generative models. Our experiments showed that IsoMax loss works as a seamless SoftMax loss drop-in replacement that significantly improves neural networks' OOD detection performance. Hence, it may be used as a baseline OOD detection approach to be combined with current or future OOD detection techniques to achieve even higher results.

Out-of-distribution (OOD) detection approaches usually present special requirements (e.g., hyperparameter validation, collection of outlier data) and produce side effects (e.g., classification accuracy drop, slower energy-inefficient inferences). We argue that these issues are a consequence of the SoftMax loss anisotropy and disagreement with the maximum entropy principle. Thus, we propose the IsoMax loss and the entropic score. The seamless drop-in replacement of the SoftMax loss by IsoMax loss requires neither additional data collection nor hyperparameter validation. The trained models do not exhibit classification accuracy drop and produce fast energy-efficient inferences. Moreover, our experiments show that training neural networks with IsoMax loss significantly improves their OOD detection performance. The IsoMax loss exhibits state-of-the-art performance under the mentioned conditions (fast energy-efficient inference, no classification accuracy drop, no collection of outlier data, and no hyperparameter validation), which we call the seamless OOD detection task. In future work, current OOD detection methods may replace the SoftMax loss with the IsoMax loss to improve their performance on the commonly studied non-seamless OOD detection problem.

Given the actual volume and speed of financial transactions, financial fraud detection systems are constantly evolving based on new computational intelligence algorithms. Therefore, transaction monitoring and analysis prevent monetary losses caused by fraudsters. Since the fraud detection process is a labor-intensive task for human auditors given the huge amount of daily transactions processed by financial services information systems. Credit card is the financial product most explored in the financial fraud detection literature, while mobile money service is becoming a popular option for payments, fraud detection for such financial product has not yet been deeply explored. Therefore, it is interesting to optimize the auditing process and test new quantitative techniques, such as deep learning, to support human auditors before double-checking a suspicious transaction. Thus, we propose an integration of adversarial autoencoders and machine learning methods to perform an objective classification among three transaction types: regular, local, and global anomaly. The integration consists of using the autoencoder's generated latent vectors as features for the supervised learning algorithms. The experiments considered different latent vector space forms concerning their dimensionality and the clusters generated by a prior Gaussian mixture. The results show that some classifiers may accept latent characteristics well, getting better or similar performance when using all the original characteristics.

This work investigates information retrieval methods to address the existing difficulties on the Preliminary Search, part of the law making process from the Brazilian Chamber of Deputies. For such, different preprocessing approaches, stemmers, language models, and BM25 variants were compared. Two legislative corpora from Chamber were used to build and validate the pipeline. All texts were converted to lowercase and had stopwords, accentuation, and punctuation removed. Words were represented by their stem combined with word unigram and bigram language models. Retrieving the bill that was originated from a specific job request, the BM25L with Savoy stemmer reached a R@20 of 0.7356. After removing queries with inconsistencies or which made reference exclusively to attachments, to other job requests, or to bills, the R@20 increased to 0.94.

Microorganisms that cause infectious diseases are defined as pathogens, as they multiply and cause tissue damage. All microorganisms isolated in culture from a location on the body should be considered potential pathogens. The infectious processes demonstrate physiological responses to the multiplication invasion of the aggressor microorganism. The disease’s development is influenced by the patient’s general health, defense mechanisms, and previous contact with the offending agent. When an infectious disease is suspected, cultures should be performed. This article uses an electronic nose to collect and analyze volatile organic compounds VOCs expelled by colonies of microorganisms. We propose signature identification of these colony odors from microorganisms using InceptionTime. The InceptionTime model is a set of models of the deep convolutional neural network, inspired by the Inception-v4 architecture. The results were excellent, with an average accuracy in the test set above 98%. The aim of our research is to propose a faster, cheaper and more accurate method of detecting these pathogens and the encouraging results of this stage encourage further research.

SigNet is a state of the art model for feature representation used for handwritten signature verification (HSV). This representation is based on a Deep Convolutional Neural Network (DCNN) and contains 2048 dimensions. When transposed to a dissimilarity space generated by the dichotomy transformation (DT), related to the writer-independent (WI) approach, these features may include redundant information. This paper investigates the presence of overfitting when using Binary Particle Swarm Optimization (BPSO) to perform the feature selection in a wrapper mode. We proposed a method based on a global validation strategy with an external archive to control overfitting during the search for the most discriminant representation. Moreover, an investigation is also carried out to evaluate the use of the selected features in a transfer learning context. The analysis is carried out on a writer-independent approach on the CEDAR, MCYT and GPDS datasets. The experimental results showed the presence of overfitting when no validation is used during the optimization process and the improvement when the global validation strategy with an external archive is used. Also, the space generated after feature selection can be used in a transfer learning context.

Most of the automatic fire alarm systems detect the fire presence through sensors like thermal, smoke, or flame. One of the new approaches to the problem is the use of images to perform the detection. The image approach is promising since it does not need specific sensors and can be easily embedded in different devices. However, besides the high performance, the computational cost of the used deep learning methods is a challenge to their deployment in portable devices. In this work, we propose a new deep learning architecture that requires fewer floating-point operations (flops) for fire recognition. Additionally, we propose a portable approach for fire recognition and the use of modern techniques such as inverted residual block, convolutions like depth-wise, and octave, to reduce the model's computational cost. The experiments show that our model keeps high accuracy while substantially reducing the number of parameters and flops. One of our models presents 71% fewer parameters than FireNet, while still presenting competitive accuracy and AUROC performance. The proposed methods are evaluated on FireNet and FiSmo datasets. The obtained results are promising for the implementation of the model in a mobile device, considering the reduced number of flops and parameters acquired.

The detection of objects considering a 6DoF pose is a common requirement to build virtual and augmented reality applications. It is usually a complex task which requires real-time processing and high precision results for adequate user experience. Recently, different deep learning techniques have been proposed to detect objects in 6DoF in RGB images. However, they rely on high complexity networks, requiring a computational power that prevents them from working on mobile devices. In this paper, we propose an approach to reduce the complexity of 6DoF detection networks while maintaining accuracy. We used Knowledge Distillation to teach portables Convolutional Neural Networks (CNN) to learn from a real-time 6DoF detection CNN. The proposed method allows real-time applications using only RGB images while decreasing the hardware requirements. We used the LINEMOD dataset to evaluate the proposed method, and the experimental results show that the proposed method reduces the memory requirement by almost 99% in comparison to the original architecture with the cost of reducing half the accuracy in one of the metrics. Code is available at https://github.com/heitorcfelix/singleshot6Dpose.

Concept drift is a change in the joint probability distribution of the problem. This term can be subdivided into two types: real drifts that affect the conditional probabilities p(y|x) or virtual drifts that affect the unconditional probability distribution p(x). Most existing work focuses on dealing with real concept drifts. However, virtual drifts can also cause degradation in predictive performance, requiring mechanisms to be tackled. Moreover, as virtual drifts frequently mean that part of the old knowledge remains useful, they require different strategies from real drifts to be effectively tackled. Motivated on this, we propose an approach called Gaussian Mixture Model for Dealing With Virtual and Real Concept Drifts (GMM-VRD), which updates and creates Gaussians to tackle virtual drifts and resets the system to deal with real drifts. The main results show that the proposed approach obtained the best results, in terms of average accuracy, in relation to the literature methods, which propose to solve that same problem. In terms of accuracy over time, the proposed approach showed lower degradation on concept drifts, which indicates that the proposed approach was efficient.

When compared to Writer-Dependent (WD) Handwritten Signature Verification, in which a model is trained for each individual writer, the Writer-Independent (WI) approach offers greater scalability, since only a single model is trained for all users from a dissimilarity space generated by the dichotomy transformation. However, many samples from the dissimilarity space are redundant and have little influence during the training of the verification model. This work investigates whether prototype selection (PS) preprocessing can be used in the space resulting from the dichotomy transformation without degrading the performance of the classifier. Furthermore, an investigation is also performed to examine the use of a WI classifier in a transfer learning scenario, i.e., where the classifier is trained in one dataset, and is used to verify signatures in other datasets. The experiments reported herein show that the use of prototype selection in the dissimilarity space allows a reduction in the complexity of the classifier without degrading its generalization performance. In addition, the results show that the WI classifier is scalable enough to be used in a transfer learning approach, with a resulting performance comparable to that of a classifier trained and tested in the same dataset. An analysis of the results obtained based on the instance hardness (IH) measure and dendrogram diagrams is performed in order to better understand the behavior of the resulting dichotomy transformation.

Semantic segmentation is an essential task in computer vision that aims to label each image pixel. Several of the actual best approaches in this context are based on deep neural networks. For example, SegNet is a deep encoder-decoder architecture approach whose results were disruptive because it is fast and performs well. However, this architecture fails to fine-delineating the edges between the objects of interest in the image. We propose some modifications in the SegNet-Basic architecture by using a post-processing segmentation layer (using Conditional Random Fields) and by transferring high resolution features combined to the decoder network. The proposed method was evaluated in the dataset CamVid. Moreover, it was compared with important variants of SegNet and showed to be able to improve the overall accuracy of SegNet-Basic by up to 17.5%.

The simplest linear combination of time series forecasters has shown a better performance than individual models. Thus, many researchers have sought to combine models for improving the forecasting process. This paper introduces a copulas-based approach (CB) for aggregating forecasters. The CB is based on the Cacoullos copula, with architecture divided into three parts: Single Modelling, Marginal Probability Distribution Computational and Joint Probability Distribution Computation. In the first part, the forecasts of individual models are obtained. In the second part, the residuals of the individuals models are calculated. In the third part, the models are combined via Cacoullos copula, based on the obtained residuals. The paper also evaluates the performance of the CB via simulated as well as financial time series (e.g. Google Stock Value). Thus, a comparative analysis is presented between CB and individual models (e.g. Artificial Neural Networks) and alternative combined forecasters (e.g. Simple Average-SA and Normal copula-CN). This study showed that the CB model produces better results when compared with the individual models, SA and CN.

Current research in the field of convolutional neural networks usually focuses on improving network accuracy, regardless of the network size and inference time. In this paper, we investigate the effects of storage space reduction in SqueezeNet as it relates to inference time when processing single test samples. In order to reduce the storage space, we suggest adjusting SqueezeNet's Fire Modules to include Depthwise Separable Convolutions (DSC). The resulting network, referred to as SqueezeNet-DSC, is compared to different convolutional neural networks such as MobileNet, AlexNet, VGG19, and the original SqueezeNet itself. When analyzing the models, we consider accuracy, the number of parameters, parameter storage size and processing time of a single test sample on CIFAR-10 and CIFAR-100 databases. The SqueezeNet-DSC exhibited a considerable size reduction (37% the size of SqueezeNet), while experiencing a loss in network accuracy of 1,07% in CIFAR-10 and 3,06% in top 1 CIFAR-100.

In this work, we investigate speaker-specific filter banks for text-independent speaker verification. The proposed method performs an heuristic search for the best filter-bank configuration using the Artificial Bee Colony (ABC) algorithm and a proper fitness function for the standard i-vectors/PLDA-based speaker verification system. Furthermore, filter-bank decorrelated amplitudes are used instead of the cepstral coefficients produced by Discrete Cosine Transform (DCT). In the experiments, the proposed method is compared to standard Mel and linear scales in both cases where the decorrelation is performed using DCT and high-pass filtering. The comparison is performed on the MIT Mobile Device Speaker Verification Corpus in a gender-dependent trial scheme. The proposed method outperformed the baseline systems in almost all the test sets for both genders. Performance gains of 4.6% and 26.0% are achieved for male and female speakers, respectively.

Software development effort estimation is the process of predicting the effort required to develop a software system. In order to improve the estimation accuracy, many different models have been proposed in the literature. Multiple classification systems represent an important field of research for machine learning. In order to estimate software development effort, this paper proposes a heterogeneous and dynamic ensemble selection model, composed by a set of regressors dynamically selected by classifiers. Along with the proposed method it is conducted an experimental analysis involving a relevant set of software effort estimation problems, which has led to better results than those achieved by classical and state of the art models previously presented.

Time series forecasting is a problem with many applications. However, in many domains, such as stock market, the underlying generating process of the time series observations may change, making forecasting models obsolete. This problem is known as Concept Drift. Approaches for time series forecasting should be able to detect and react to concept drift in a timely manner, so that the forecasting model can be updated as soon as possible. Despite the fact that the concept drift problem is well investigated in the literature, little effort has been made to solve this problem for time series forecasting so far. This work proposes two novel methods for dealing with the time series forecasting problem in the presence of concept drift. The proposed methods benefit from the Particle Swarm Optimization (PSO) technique to detect and react to concept drifts in the time series data stream. It is expected that the use of collective intelligence of PSO makes the proposed method more robust to false positive drift detections while maintaining a low error rate on the forecasting task. Experiments show that the methods achieved competitive results in comparison to state-of-the-art methods.

Efforts have been made in financial markets to deal with price movement predicting. Recent studies have shown that the market can be outperformed by methodologies with the aid of science. In other words, it has been shown that methods based on computational intelligence can be more profitable than a buy-and-hold strategy. This paper proposes a probabilistic and dynamic chart pattern recognition hybrid system, by using soft computing techniques, such as Linearized Fuzzy C-Medoids (LFCMdd), Perceptually Important Points (PIP), Dynamic Time Warping (DTW) and Probabilistic Support Vector Machines (PSVM), to predict three important foreign exchange rates. Also, comparisons are performed against trading systems built with the use of technical indicators, optimized by Genetic Algorithms (GA). The experiments were performed with the sliding window strategy. The results showed that the proposed methodology outperformed the GA-based methodology, with mean return (ROI) of 277.6% against 20.96% for the GA-based.

A time series is a sequence of observations collected over fixed sampling intervals. Several real-world dynamic processes can be modeled as a time series, such as stock price movements, exchange rates, temperatures, among others. As a special kind of data stream, a time series may present concept drift, which affects negatively time series analysis and forecasting. Explicit drift detection methods based on monitoring the time series features may provide a better understanding of how concepts evolve over time than methods based on monitoring the forecasting error of a base predictor. In this paper, we propose an online explicit drift detection method that identifies concept drifts in time series by monitoring time series features, called Feature Extraction for Explicit Concept Drift Detection (FEDD). Computational experiments showed that FEDD performed better than error-based approaches in several linear and nonlinear artificial time series with abrupt and gradual concept drifts.

Time series combined forecasters have been superior to the respective single models in statistical terms. In this way, the linear combination functions, e.g. the simple average (SA) and the minimal variance (MV) approaches, have been the main alternatives for aggregation in the literature. In this work, it is proposed a copulas-based method for combining biased single models. Copulas are multivariate functions that operate on marginal probability distributions, allowing one to model the forecasters errors and then the dependence among them: a typical divide-and-conquer framework that can result in nonlinear accurate combined forecasters. The performance of the copulas-based combination method is assessed by means of a comparison with SA and MV models, based on two financial time series.

Several approaches have been investigated to develop models able to solve forecasting problems. However, a limitation arises in the particular case of daily-frequency financial forecasting and is called the random walk dilemma (RWD). In this context, the concept of time phase adjustment can be included in forecasting models to overcome such a drawback. But the evolution of trading systems has increased the frequency for performing operations in the stock market for fractions of seconds, which requires the analysis of high-frequency financial time series. Thus, this work proposes a model, called the increasing decreasing linear neuron (IDLN), to forecast high-frequency financial time series from the Brazilian stock market. Furthermore, an evolutionary covariance-based method with automatic time phase adjustment is presented for the design of the proposed model, and the obtained results overcame those obtained by classical forecasting models in the literature.

A wide number of sophisticated models have been proposed in the literature to solve prediction problems. However, a drawback arises in the particular case of financial prediction problems and called the random walk dilemma (RWD). In this context, the concept of time phase adjustment can be used to overcome the problem for daily-frequency financial time series. However, the fast evolution of trading platforms increased the frequency for performing operations in the stock market for fractions of seconds, which makes the analysis of high-frequency financial time series very important in this current scenario. In this way, this paper presents a model, called the increasing decreasing linear neuron (IDLN), to predict high-frequency financial time series from the Brazilian stock market. Besides, a descending gradient-based method with automatic time phase adjustment is presented for the design of the proposed model, and the obtained results overcame those obtained by established prediction models in the literature.

Financial markets are very important to the economical and social organization of modern society. Due to they importance, several researchers have investigated how to predict future market movements by using both statistical and soft computing methods based on historical time series data. However, as a typical data stream, financial time series frequently present concept drift, which is a change in the relationship between input data and the target variable over time. The concept drift phenomenon affects negatively the forecasting accuracy since the learned model becomes outdated after a change in the current concept. In this paper we investigate how to handle concept drift in financial time series prediction in order to improve the forecasting accuracy. Two explicit drift detector mechanisms, namely the Drift Detection Mechanism (DDM) and the Exponentially Weighted Moving Average for Concept Drift Detection Mechanism (ECDD), were investigated. The main contribution of this work is an approach that combines Online Sequential Extreme Learning Machines (OS-ELM) with explicit drift detection, in which the OS-ELM updates the decision model just in the presence of concept drift in data. Experimental results showed that the use of drift detection was able to speed up the prediction time of OS-ELM maintaining equivalent accuracy.

Particle Swarm Optimization algorithm (PSO), when applied to problems with continuous variables, presents results with better quality at a lower computational cost when compared to the Genetic Algorithm (GA). Thus, the PSO becomes a very useful method to be applied with investment strategies used to optimize profit from operations made in the stock market, since investors seek quick and profitable results for their decision making. In this context, the Symbolic Aggregate Approximation (SAX) and Piecewise Aggregate Approximation (PAA) are time series representation techniques that, when used with optimization algorithms like PSO or GA, can help investors discover hidden and relevant patterns in financial time series data. The SAX uses discrete variables to represent their values, whereas PAA uses continuous values. Thus, this paper proposes the PAA-PSO technique, which combines the PAA with the optimization of PSO for discovery of patterns that will be used with a formulated investment strategy in order to maximize the profit from operations made in the stock market. Experiments that compare the proposed method to the results of the SAX-GA technique, which combines the techniques of SAX and GA in their investment strategy, are reported.

Considerable effort has been made by researchers from various areas of science to forecast financial time series such as stock market and foreign exchange market. Recent studies have shown that the market can be outperformed by trading systems built with soft computing techniques. This paper aims to compare different trading systems based on support vector regression (SVR), growing hierarchical self-organizing maps (GHSOM) and genetic algorithms (G A) when tested against nine currency pairs of the foreign exchange market (Forex). The experiments were performed using the sliding window strategy. The results showed that the GA-based trading systems outperformed the SVR+GHSOM model when evaluated by four performance metrics, including an statistical test.

Financial markets are very important to the economical and social organization of modern society. In this kind of market, the success of an investor depends on the quality of the information he uses to trade in the market, and on how fast he is able to take decisions. In the literature, several statistical and soft computing mechanisms have been proposed in order to support investors decision in the financial market. In this work we propose an autonomous trader agent that is able to compute technical indicators of the stock market and take decisions on buying or selling stocks. Our trader agent is based on a single hidden layer feedforward (SLFN) ensemble trained with online sequential extreme learning machine (OS-ELM), a variant of ELM that is able to learn data one-by-one and dynamically accommodate changes in the market. In addition, we propose a set of trading rules that guides the trader agent in order to improve the potential profit. Experimental results on real dataset from Brazilian stock market showed that our proposed trader agent based on OS-ELM ensemble is able to increase the financial gain when compared with other approaches proposed in literature.

As has been shown by the recent literature, machine learning techniques are important tools for diagnosing a number of diseases. Hospitals and medical clinics store a large amount of data with respect to the treatment of their patients. However, rarely an analysis of these data is conducted in order to extract intrinsic information for modeling a specific problem. This work presents an analysis of medical data aimed at determining whether or not patients are cardiac. To this end, raw data was collected and preprocessed at a Brazilian local hospital in order to build a new dataset containing only non-invasive information of children with heart murmur symptoms. The gathered data contain information, such as height, weight, gender and birthday date. The collected data was shown to be very imbalanced. Due to this imbalance, we employ the One-class Classification (OCC) paradigm to solve the problem by experimenting five methods; including the FBDOCC, that we proposed in a previous paper. Furthermore, two additional datasets were experimented in order to assess effectiveness of One-Class classifiers on the domain of heart disease detection. The overall results show that the FBDOCC succeeded in this task, yielding, statistically, the best performance for the gathered dataset as well as the other two heart disease datasets.

In this paper we evaluate the combination of Extreme Learning Machine (ELM) and Support Vector Regression (SVR) with a Kalman filter regression model for financial time series forecasting. We also compare the forecast performance with a set of linear regression combination methods. The application of the traditional Kalman Filter for the statistical arbitrage strategy improves the statistical performance of ELM and SVR individual forecasts. The accuracy of the models is statistically tested and an investigation is performed to confirm the impact of the forecasts combination in terms of annualized returns and volatility.

The dilation-erosion-linear perceptron (DELP) is a morphological-linear model based on fundamentals of mathematical morphology (MM). Its design is a gradient-based learning process using ideas from the backpropagation (BP) algorithm. However, a drawback arises from the gradient estimation of morphological operators, because they are not differentiable of usual way. In this sense, this paper presents an evolutionary learning process, using the covariance matrix adaptation evolutionary strategy (CMAES), to design the DELP model. Furthermore, we conduct an experimental analysis using a relevant set of binary classification problems, and the obtained results are discussed and compared to results found using the DELP model with its classical learning process.

There are many researches on forecasting time series for building trading systems for financial markets. Some of these studies have shown that it is possible to obtain satisfactory results, thereby contradicting the theory of Efficient Markets Hypothesis (EMH) that suggests that prices are randomly generated over time. This paper proposes an intelligent system based on historical closing prices that uses technical analysis, the Artificial Bee Colony Algorithm (ABC), a selection of past values (lags), nearest neighbor classification (k-NN) and its variation, the Adaptative Classification and Nearest Neighbor (A-k-NN). A very important step for time series prediction is the correct selection of the past observations (lags). Our method uses this strategy since it uses the k-NN and A-k-NN to decide on the buy and seIl points, combined with the ABC algorithm which is used to search for the best parameter settings of system and a good set of lags. This paper compares the results obtained by the proposed method with the buy and hold strategy and with other work that performed similar experiments with the same trading model and the same stocks. The key measure for performance comparison is the profitability in the analyzed period. The proposed method generates much larger profits compared to the other method and to the buy and hold strategy. Our method outperforms the other methods in thirteen out of the fifteen stocks tested, minimizing the risk of market ex pos ure.

In this paper we investigate the statistical and economic performance for statistical arbitrage strategy using Extreme Learning Machine (ELM) and Support Vector Regression (SVR) models, and their forecast combination through four linear combination models. The application of the traditional Kalman Filter for the statistical arbitrage strategy improves the statistical performance of ELM and SVR individual forecasts. It is presented evidence that the financial performance for most of cointegrated pairs can be improved by at least one linear combination technique.

Machine learning techniques are an important tool for diagnosing a number of diseases, as has been shown by the recent literature. Hospitals and medical clinics have a huge amount of data about the treatment of their patients, however, rarely analysis of these data is performed in order to extract intrinsic information aimed at modeling a specific problem. This work presents an analysis of medical data aimed at determining whether children patients are cardiac or not. To this end, raw data was collected at a Brazilian local hospital to be preprocessed in order to build the classification models. Only non invasive information were used, such as height, weight, gender and birthday date to create another set of derived variables such as BMI (Body Mass Index) to support the classification phase. However, the collected data was shown to be very imbalanced. Aimed at treat this problem, many tecniques were employed and one new approach was proposed. The results shown that the proposed approach outperforms the other methods in three out of four evaluation metrics.

Considerable effort has been made by researchers from various areas of science to forecast financial time series such as stock market and foreign exchange market (Forex). Recent studies have shown that the market can be outperformed by trading systems built with computational intelligence techniques. This study applies the Genetic Algorithm (GA) technique to optimize technical indicators parameters in order to maximize profit in the nine most tradable foreign exchange rates. Fifteen trading systems were created by combining four technical indicators optimized by the GA. It is then compared to an SVR+GHSOM model trading system and an analysis is performed to assess the most adaptable model in a period of international economic crisis. We report in the experiments that the GA model was far superior compared to the SVR+GHSOM model in the test period. The comparison considered performance measures such as profitability (ROI) and the maximum draw down (MD). The experiments have also shown that it is possible to increase profit by adjusting the risk parameter (lots size), at the expense of increasing the risk.

There are many researches aimed to predict times series of various financial markets. Some of these papers have shown that it is possible to obtain satisfactory results, thereby contradicting the theory that financial time series follow a random walk model. This study applies an architecture based on two stages for trading with two of the most traded foreign exchange rates (forex), the EUR/USD and GBP/USD. It also proposes a trading system to evaluate the model under a financial perspective, both in terms of profitability and risk, and to compare the application of the model in different timeframes (daily or intraday). The architecture consists of a GHSOM network, whose goal is to divide the dataset into regions with similar statistical distribution in order to circumvent the problem of nonstationarity, and a support vector regression machine (SVR), to make forecasts for the regions defined by GHSOM. We report on experiments that the SVR+GHSOM architecture performance is far superior compared to a model based solely on SVR. The comparison considered performance measures such as profitability (ROI) and the maximum drawdown (MD) and has shown that the best results are obtained in daily timeframe. The experiments have also shown that it is possible to increase profit by adjusting the risk parameter (number of lots), at the expense of increasing the risk. Furthermore, the proposed model proved to be much more profitable than a buy-and-hold model using the same time series (EUR/USD and GBP/USD); it also outperformed buy-and-hold with the Dow Jones in the same period.

In this work we present an evolutionary learning process using the covariance matrix adaptation evolutionary strategy (CMAES) to design the dilation-erosion perceptron (DEP) for stock market indices forecasting. Also, we have included an automatic phase fix procedure (APFP) into proposed learning process to eliminate time phase distortions observed in some forecasting problems. The main advantage of the DEP model designed by our learning process, apart from its higher forecasting performance, is do not request any methodology to overcome the nondifferentiability of morphological operators needed into classical gradient-based learning process of the DEP model. Besides, we present an experimental analysis using two stock market indices, where five well-known performance metrics and an evaluation function are used to assess forecasting performance.

In this paper we present a gradient-based morphological approach to solve the software development cost estimation (SDCE) problem. The proposed approach consists of a dilation-erosion perceptron (DEP) trained by a gradient steepest descent method using the back propagation (BP) algorithm and a systematic approach to overcome the problem of nondifferentiability of morphological operators. Furthermore, we compare the proposed approach with other neural and statistical models using five complex SDCE problems.

The dilation-erosion perceptron (DEP) is a class of hybrid artificial neurons based on framework of mathematical morphology (MM) with algebraic foundations in the complete lattice theory (CLT). A drawback arises from the gradient estimation of dilation and erosion operators into classical gradient-based learning process of the DEP model, since they are not differentiable of usual way. In this sense, we present a differential evolutionary learning process, called DEP(MDE), using a modified differential evolution (MDE) to design the DEP model for air pressure forecasting. Also, we have included an additional step into learning process, called automatic phase fix procedure (APFP), to eliminate time phase distortions observed in some forecasting problems. Furthermore, an experimental analysis is presented using two complex time series, where five well-known performance metrics and an evaluation function are used to assess forecasting performance.

The Dilation-erosion perceptron (DEP) is considered a good forecasting model, whose foundations are based on mathematical morphology (MM) and complete lattice theory (CLT). However, a drawback arises from the gradient estimation of morphological operators into classical gradient-based learning process, since they are not differentiable of usual way. In this sense, this work presents an evolutionary learning process, called DEP(MGA), using a modified genetic algorithm (MGA) to design the DEP model for weather forecasting. In addition, we have included an automatic phase fix procedure (APFP) into the proposed learning process to eliminate time phase distortions observed in some temporal phenomena. At the end, an experimental analysis is presented using two complex time series, where five well-known performance metrics and an evaluation function are used to assess forecasting performance.

Find software defects is a complex and slow task which consumes most of the development budgets. In order to try reducing the cost of test activities, many researches have used machine learning to predict whether a module is defect-prone or not. Defect detection is a cost-sensitive task whereby a misclassification is more costly than a correct classification. Yet, most of the researches do not consider classification costs in the prediction models. This paper introduces an empirical method based in a COCOMO (COnstructive COst MOdel) that aims to assess the cost of each classifier decision. This method creates a cost matrix that is used in conjunction with a threshold-moving approach in a ROC (Receiver Operating Characteristic) curve to select the best operating point regarding cost. Public data sets from NASA (National Aeronautics and Space Administration) IV&V (Independent Verification & Validation) Facility Metrics Data Program (MDP) are used to train the classifiers and to provide some development effort information. The experiments are carried out through a methodology that complies with validation and reproducibility requirements. The experimental results have shown that the proposed method is efficient and allows the interpretation of the classifier performance in terms of tangible cost values.

One-class classification is an important problem with applications in several different areas such as outlier detection and machine monitoring. In this paper we propose a novel method for one-class classification which also implements prototype reduction. The main feature of the proposed method is to analyze every limit of all the feature dimensions to find the true border which describes the normal class. To this end, the proposed method simulates the novelty class by creating artificial prototypes outside the normal description. The method is able to describe data distributions with complex shapes. Aiming to assess the proposed method, we carried out experiments with synthetic and real datasets to compare it with the Support Vector Domain Description (SVDD), kMeansDD, ParzenDD and kNNDD methods. The experimental results show that our one-class classification approach outperformed the other methods in terms of the area under the receiver operating characteristic (ROC) curve in three out of six data sets. The results also show that the proposed method remarkably outperformed the SVDD regarding training time and reduction of prototypes.

Scheduling is an important tool for optimizing the performance of parallel systems. It aims at reducing the completion time of parallel applications by properly allocating the tasks to the processors. This work proposes a novel scheduling algorithm to parallelize tasks with dependence restrictions. The communication costs between processors and computer architecture are parameters of the proposed algorithm, which explores the trade off between process execution time and communication costs between processes to optimize the system's overall performance. The paper conducts an experiment to compare the performance of the proposed algorithm against six other scheduling algorithms. The experiment considered several execution scenarios. Although our algorithm does not present the best performance in any of the execution scenarios, it produces the best average execution time for the scenarios studied.

Real problem simulations involving physic phenomena can demand too much execution time. To improve the performance of these simulations it is necessary to have an approach to parallelize the processes that compose the simulation. MPhyScaS (Multi-Physics and Multi-Scale Solver Environment) is an environment dedicated to the automatic development of simulators. Each MPhyScaS simulation demands a great amount of time. To parallelize MPhyScaS simulations, the approach used should define a hierarchical parallel structure. The aim of the work herein presented is to improve the performance of clusters in the processing of MPhyScaS simulations which are composed by a set of dependent tasks by scheduling them. The presented model is based on Genetic Algorithms (GA) to schedule the parallel tasks following MPhyScaS architecture dependence restrictions. The communication between processors must also be considered in this scheduling. Therefore, a trade off must be found between the execution of processes and the time necessary for these processes to communicate with each other.

In this paper we present a methodology, called covariance matrix adaptation based evolutionary (CMAbE), to solve the financial time series forecasting problem. The proposed methodology consists of a hybrid model composed of multilayer perceptrons (MLPs) combined with the Covariance Matrix Adaptation Evolution Strategy (CMAES), which determines the most fitted time lags to characterize the time series phenomenon, as well as searches for the best architecture, parameters and training algorithm of MLP networks. An experimental analysis is conducted with the proposed methodology through two real world financial time series, and the obtained results are discussed and compared to results found with recently methods presented in literature.

This paper presents a hybrid intelligent method to design Morphological-Rank-Linear (MRL) perceptrons to solve the Software Development Cost Estimation (SDCE) problem. The proposed method uses a modified genetic algorithm (MGA) to determine the best particular features to improve the MRL perceptron performance, as well as its initial parameters. Furthermore, for each individual of MGA, a gradient steepest descent method is used to optimize the MRL perceptron parameters supplied by MGA. An experimental analysis is conducted with the proposed method using the Desharnais and Cocomo databases. In the experiments, two relevant performance metrics and a fitness function are used to assess the performance of the proposed method. The results obtained are compared to methods recently presented in literature.

In this work a quantum-inspired hybrid methodology is proposed to overcome the random walk dilemma for financial time series prediction. It consists of a hybrid model composed of a Qubit Multilayer Perceptron (QuMLP) with a Quantum-Inspired Evolutionary Algorithm (QIEA), which searches for the best particular time lags able to characterize the time series phenomenon, as well as to evolve the complete QuMLP architecture and parameters. Each individual of the QIEA population is adjusted by the Complex Back-Propagation (CBP) algorithm to further improve the QuMLP parameters supplied by the QIEA. After the prediction model search procedure, it uses a behavioral statistical test and a phase fix procedure to adjust time phase distortions that appear in financial time series. An experimental analysis is conducted with the proposed methodology through four real world financial time series, and the obtained results are discussed and compared to results found with Multilayer Perceptiron (MPL) networks and the previously introduced Morphological-Rank-Linear Time-lag Added Evolutionary Forecasting (MRLTAEF) method.

This paper introduces a novel instance-based one-class classification method for novelty detection in time series based on its states transition. The main feature of our work is to generate an efficient method which automatically finds the parameters (whose yields the best model) according with the quality of the discovered time series states and the validation error. This method involves clustering and reducing the number of samples in a training dataset which does not contain novelty samples. Experiments carried out using three real-world time series show that the proposed method is able to build models with a reduced number of stored prototypes. The results obtained by our method were compared with the results of the SAX and both methods have successfully detected the novelties, however, the parameters which resulted in the best SAX model were achieved without validation phase (i.e. analyzing the results obtained for the test set).

This paper proposes the Covariance Matrix Adaptation based Evolutionary (CMAbE) methodology to overcome the random walk dilemma, characterized by one step delay regarding the real time series values, adjusting time phase distortions in the financial time series forecasting problem. The proposed CMAbE methodology consists of a hybrid model composed of the MultiLayer Perceptron (MLP) networks and the Covariance Matrix Adaptation Evolution Strategy (CMAES), which searches for the best particular time lags to optimally describe the time series phenomenon, as well for the best architecture, parameters and training algorithm of MLP networks. An experimental analysis is conducted with the proposed methodology through four real world financial time series, and the obtained results are discussed and compared to results found with recently methods presented in literature.

Real problem simulations involving physic phenomena can demand too much execution time. To improve the performance of these simulations it is necessary to have an approach to parallelize the processes that compose the simulation. MPhyScaS (Multi-Physics and Multi-Scale Solver Environment) is an environment dedicated to the automatic development of simulators. Each MPhyScaS simulation demands a great amount of time. To parallelize MPhyScaS simulations, the approach used should define a hierarchical parallel structure. The aim of the work herein presented is to identify parallel jobs and dependent ones. The presented model is based on Coloured Petri Nets (CPN). This information will be input to a scheduling algorithm.

This work presents a Morphological-Rank-Linear approach to solve the problem of Software Development Cost Estimation (SDCE). It consists of a hybrid morphological model, which is a linear combination between a Morphological-Rank (MR) operator (nonlinear) and a Finite Impulse Response (FIR) operator (linear), referred to as Morphological-Rank-Linear (MRL) filter. A gradient steepest descent method to adjust the MRL filter parameters (learning process), using the Least Mean Squares (LMS) algorithm, and a systematic approach to overcome the problem of nondifferentiability of the morphological-rank operator are used to improve the numerical robustness of training algorithm. Furthermore, an experimental analysis is conducted with the proposed approach using the well-known NASA database. In the experiments, two relevant performance metrics and an evaluation function are used to assess the performance of the proposed approach. The results obtained are compared to models recently presented in literature.

This paper presents the results of method designed to predict price trends in the stock market. Our first and foremost objective is to study the feasibility of the practical use of an intelligent prediction system exclusively based on the history of daily stock closing prices and volumes. To this end we propose a technique that consists of a combination of a nearest neighbor classifier and some well known tools of technical analysis, namely, stop loss, stop gain and RSI filter. For assessing the potential use of the proposed method in practice we compared the results obtained to the results that would be obtained by adopting a buy-and-hold strategy. The key performance measure in this comparison was profitability. The proposed method was shown to generate considerable higher profits than buy-and-hold for most of the companies, with few buy operations generated and, consequently, minimizing the risk of market exposure.

A novel method is proposed herein for handwritten digit segmentation in historical document images. It is based on one-class classifiers, which are used to distinguish isolated characters from touching characters. In contrast to other techniques based on feed forward neural networks, the proposed method does not require negative data in the training phase. Three methods for feature extraction and five one class classifiers are considered and have their performance compared. Experimental results on a data set of handwritten digits extracted from a collection of historical documents show the effectiveness of the proposed method.

There are striking disparities in the prevalence of dental disease by income. Poor children suffer twice as much dental caries as their more affluent peers, but are less likely to receive treatment. This paper presents an experimental study of the application of machine learning methods to the problem of caries prediction. For this paper a data set collected from interviews with children under five years of age, in 2006, in Recife, the capital of Pernambuco, a state in northeast Brazil, was built. Four different data mining techniques were applied to this problem and their results were confronted in terms of the classification error and area under the ROC curve (AUC). Results showed that the MLP neural network classifier out performed the other machine learning methods employed in the experiments, followed by the support vector machine (SVM) predictor. In addition, the results also show that some rules (extracted by decision tress) may be useful for understanding the most important factors that influence the occurrence of caries in children.

Routing is very important for computer networks because it is one of the main factors that influences network performance. In this paper, we propose an improved intelligent method for routing based on Hopfield Neural Networks (HANN), which uses a discrete equation and the Particle Swarm Optimization (PSO) technique to optimize the HNN parameters. The fitness function for the PSO algorithm used here is a combination of the number of iterations for convergence and the percentage error when the HNN method tries to find the best path in a communication network. The simulation results show that PSO is a reliable approach to optimize the Hopfield network for routing in computer networks, since this method results in fast convergence and produces accurate results.

The precision and reliability of the estimation of the effort of software projects is very important for the competitiveness of software companies. Good estimates play a very important role in the management of software projects. Most methods proposed for effort estimation, including methods based on machine learning, provide only an estimate of the effort for a novel project. In this paper we introduce a method based on machine learning which gives the estimation of the effort together with a confidence interval for it. In our method, we propose to employ robust confidence intervals, which do not depend on the form of probability distribution of the errors in the training set. We report on a number of experiments using two datasets aimed to compare machine learning techniques for software effort estimation and to show that robust confidence intervals can be successfully built.

Efficient routing algorithms are very important for the operation of communication networks, including the Internet. This article proposes a novel hybrid intelligent method for routing which combines Hopfield neural networks (HNN) and simulated annealing (SA). The proposed method introduces a modified version of the discrete-time equation used by Bastos-Filho et al [1]. The novel version of the equation aims to improve the HNN convergence, thereby decreasing the computation cost. In our method, the SA algorithm is used to obtain the optimal parameters of the HNN. Simulations reported in this paper shows that the proposed method outperforms the method of Bastos-Filho et al [1], by computing routes using smaller number of iterations and smaller error.

Software reuse is essential for improving the productivity and quality of software projects. One of the key issues to promote the adoption of software reuse in companies is the development of effective repositories of software components. It is also very important to have good methods for searching and retrieval of the components. Clustering techniques can help by providing a visualization of the repository of software components as well as in helping to refine the searches by grouping together similar components. In this paper we quantitatively compare two clustering techniques, namely, self-organizing maps (SOM) and growing hierarquical SOM (GHSOM) for clustering a repository of classes from a Java API for building mobile systems. The performance measure was the quantization error. The simulations have shown that GHSOM outperforms SOM in these tasks. GHSOM is more suitable for this task because it is a constructive technique, which is an advantage in tackling the growth of the repository of software components.

The precision and reliability of the estimation of the effort of software projects is very important for the competitiveness of software companies. Good estimates play a very important role in the management of software projects. Most methods proposed for effort estimation, including methods based on machine learning, provide only an estimate of the effort for a novel project. In this paper we introduce a method based on machine learning which gives the estimation of the effort together with a confidence interval for it. In our method, we propose to employ robust confidence intervals, which do not depend on the form of probability distribution of the errors in the training set. We report on a number of experiments using two datasets aimed to compare machine learning techniques for software effort estimation and to show that robust confidence intervals for the effort estimation can be successfully built.

This paper presents a new algorithm for binarization of historical document images. It is adjusted to deal with the particular case of documents which are written in both sides of the paper. The digitization of these documents in general brings to the digital image this back-to-front interference. In order to deal with it, a thresholding algorithm is proposed based on color quantization by genetic algorithms and image fidelity analysis. The method achieved better results than other well-known algorithms.

Software reuse is essential for improving the productivity and quality of software projects. One of the key issues to promote the adoption of software reuse in companies is the development of effective repositories of software components. It is also very important to have good methods for searching and retrieval of the components. Clustering techniques can help by providing a visualization of the repository of software components as well as in helping to refine the searches by grouping together similar components. In this paper we quantitatively compare two clustering techniques, namely, self-organizing maps (SOM) and growing hierarchical SOM (GHSOM) for clustering a repository of classes from a Java API for building mobile systems. The performance measure was the quantization error. The simulations have shown that GHSOM outperforms SOM in these tasks. GHSOM is more suitable for this task because it is a constructive technique, which is an advantage in tackling the growth of the repository of software components.

Much of the current research in software defect prediction focuses on building classifiers to predict only whether a software module is fault-prone or not. Using these techniques, the effort to test the software is directed at modules that are labelled as fault-prone by the classifier. This paper introduces a novel algorithm based on constructive RBF neural networks aimed at predicting the probability of errors in fault-prone modules; it is called RBF-DDA with Probabilistic Outputs and is an extension of RBF-DDA neural networks. The advantage of our method is that we can inform the test team of the probability of defect in a module, instead of indicating only if the module is fault-prone or not. Experiments carried out with static code measures from well-known software defect datasets from NASA show the effectiveness of the proposed method. We also compared the performance of the proposed method in software defect prediction with kNN and two of its variants, the S-POC-NN and R-POC-NN. The experimental results showed that the proposed method outperforms both S-POC-NN and R-POC-NN and that it is equivalent to kNN in terms of performance with the advantage of producing less complex classifiers.

This paper proposes and investigates the use of bagging predictors to improve performance of regression methods for estimation of the effort to develop software projects. We have applied bagging to M5P/regression trees, M5P/model trees, multi-layer perceptron (MLP), linear regression and support vector regression (SVR). This article reports on the influence of bagging on the performance of each of these regression methods in the estimation of the effort of software projects. Experiments carried out using a dataset of software projects from NASA show that bagging is able to significantly improve performance of regression methods in this task. Moreover, we show that bagging with M5P/model trees considerably outperforms previous results reported in the literature obtained by both linear regression and RBF networks. It is also shown that bagging with M5P/model trees obtains results comparable to those of SVR, with the advantage of producing more interpretable results.

One-class classification is an important problem with applications in several different areas such as novelty detection, outlier detection and machine monitoring. In this paper we propose a novel method for one-class classification, referred to as NNDDSRM. It is based on the principle of structural risk minimization and the nearest neighbor data description (NNDD) method. Experiments carried out using both artificial and real-world datasets show that the proposed method is able to significantly reduce the number of stored prototypes in comparison to NNDD. The experimental results also show that the proposed method outperformed NNDD - in terms of the area under the receiver operating characteristic (ROC) curve - on four of the five datasets considered in the experiments and had a similar performance on the remaining one.

This article proposes a new hybrid technique to optimize an intelligent routing algorithm in telecommunications based on Hopfield Neural Networks (HNN) and Simulated Annealing (SA). The SA obtains the optimal parameters for Hopfield Neural Networks. The optimization was carried out considering as performance criterion a combination of (1) the error of the HNN routing algorithm as applied to find the shortest path between two nodes, and (2) the number of iterations employed by the HNN algorithm to find the shortest path. By using this approach better results have been achieved, with lower number of iterations and smaller error rates.

The Dynamic Decay Adjustment (DDA) algorithm is a fast constructive algorithm for training RBF neural networks. This paper proposes a method for improving RBF-DDA generalization performance by adjusting the weights of the connections between hidden and output units. The method proposed here has been evaluated on three optical character recognition datasets from the UCI repository. The results show that the proposed method considerably improves performance of RBF-DDA in these tasks without increasing the size of the networks. The results are compared to MLP, k-NN, AdaBoost and SVM results reported in the literature. It is shown that the proposed method outperforms MLP and AdaBoost and obtains results comparable to k-NN and SVM on these datasets.

The dynamic decay adjustment (DDA) algorithm is a fast constructive algorithm for training RBF neural networks (RBFNs) and probabilistic neural networks (PNNs). The algorithm has two parameters, namely, theta(+) and theta(-). The papers which introduced DDA argued that those parameters would not heavily influence classification performance and therefore they recommended using always the default values of these parameters. In contrast, this paper shows that smaller values of parameter theta(-) can, for a considerable number of datasets, result in strong improvement in generalization performance. The experiments described here were carried out using twenty benchmark classification datasets from both Proben1 and the UCI repositories. The results show that for eleven of the datasets, the parameter theta(-) strongly influenced classification performance. The influence of theta(-) was also noticeable, although much less, on six of the datasets considered. This paper also compares the performance of RBF-DDA with theta(-) selection with both AdaBoost and Support Vector Machines (SVMs).

The dynamic decay adjustment (DDA) algorithm is a fast constructive algorithm for training RBF and PNN neural networks. The algorithm has two parameters, namely, /spl theta//sup +/ and /spl theta//sup -/. The papers which introduced DDA argued that those parameters would not heavily influence classification performance and therefore they recommended using always the default values of these parameters. In contrast, this paper shows that smaller values of parameter /spl theta/ can, for a considerable number of datasets, result in remarkable improvement in generalization performance.

Novelty detection in time series is an important problem with application in different domains. such as machine failure detection, fraud detection and auditing. We have previously proposed a method for time series novelty detection based on classification of time series windows by RBF-DDA neural networks. The paper proposes a method to be used in conjunction with this time series novelty detection method whose aim is to improve performance by adequately selecting the window size and the RBF-DDA parameter values. The method was evaluated on six real-world time series and the results obtained show that it greatly improves novelty detection performance.