Mach. Learn. Knowl. Extr., Volume 1, Issue 3 (September 2019) – 15 articles

The modeling of time to event data is an important topic with many applications in diverse areas. The collective of methods to analyze such data are called survival analysis, event history analysis or duration analysis. Survival analysis is widely applicable because the definition of an ’event’ can be manifold and examples include death, graduation, purchase or bankruptcy. Hence, application areas range from medicine and sociology to marketing and economics. In this paper, we review the theoretical basics of survival analysis including estimators for survival and hazard functions. We discuss the Cox Proportional Hazard Model in detail and also approaches for testing the proportional hazard (PH) assumption. Furthermore, we discuss stratified Cox models for cases when the PH assumption does not hold. Our discussion is complemented with a worked example using the statistical programming language R to enable the practical application of the methodology. Full article

Several suppliers of oil and gas (O & G) equipment and services have reported the necessity of making frequent resources planning adjustments due to the variability of demand, which originates in unbalanced production levels. The occurrence of these specific problems for the suppliers and operators is often related to the bullwhip effect. For studying such a problem, a research proposal is herein presented. Studying the bullwhip effect in the O & G industry requires collecting data from different levels of the supply chain, namely: services, upstream and midstream suppliers, and downstream clients. The first phase of the proposed research consists of gathering the available production and financial data. A second phase will be the statistical treatment of the data in order to evaluate the importance of the bullwhip effect in the oil and gas industry. The third phase of the program involves applying artificial neural networks (ANN) to forecast the demand. At this stage, ANN based on different training methods will be used. Further on, the attained mathematical model will be used to simulate the effects of demand fluctuations and assess the bullwhip effect in an oil and gas supply chain. Full article

Prediction is a common machine learning (ML) technique used on building energy consumption data. This process is valuable for anomaly detection, load profile-based building control and measurement and verification procedures. Hundreds of building energy prediction techniques have been developed over the last three decades, yet there is still no consensus on which techniques are the most effective for various building types. In addition, many of the techniques developed are not publicly available to the general research community. This paper outlines a library of open-source regression techniques from the Scikit-Learn Python library and describes the process of applying them to open hourly electrical meter data from 482 non-residential buildings from the Building Data Genome Project. The results illustrate that there are several techniques, notably decision tree-based models, that perform well on two-thirds of the total cohort of buildings. However, over one-third of the buildings, specifically primary schools, performed poorly. This example implementation shows that there is no one size-fits-all modeling solution and that various types of temporal behavior are difficult to capture using machine learning. An analysis of the generalizability of the models tested motivates the need for the application of future techniques to a board range of building types and behaviors. The importance of this type of scalability analysis is discussed in the context of the growth of energy meter and other Internet-of-Things (IoT) data streams in the built environment. This framework is designed to be an example baseline implementation for other building energy data prediction methods as applied to a larger population of buildings. For reproducibility, the entire code base and data sets are found on Github. Full article

In real world applications, binary classification is often affected by imbalanced classes. In this paper, a new methodology to solve the class imbalance problem that occurs in image classification is proposed. A digital image is described through a novel vector-based representation called Kernel Graph Embedding on Attributed Relational Scale-Invariant Feature Transform-based Regions Graph (KGEARSRG). A classification stage using a procedure based on support vector machines (SVMs) is organized. Methodology is evaluated through a series of experiments performed on art painting dataset images, affected by varying imbalance percentages. Experimental results show that the proposed approach consistently outperforms the competitors. Full article

Statistical hypothesis testing is among the most misunderstood quantitative analysis methods from data science. Despite its seeming simplicity, it has complex interdependencies between its procedural components. In this paper, we discuss the underlying logic behind statistical hypothesis testing, the formal meaning of its components and their connections. Our presentation is applicable to all statistical hypothesis tests as generic backbone and, hence, useful across all application domains in data science and artificial intelligence. Full article

In recent years the emergence of social media has become more prominent than ever. Social networking has become the de facto tool used by people all around the world for information discovery. Consequently, the importance of recommendations in a social network setting has urgently emerged, but unfortunately, many methods that have been proposed in order to provide recommendations in social networks cannot produce scalable solutions, and in many cases are complex and difficult to replicate unless the source code of their implementation has been made publicly available. However, as the user base of social networks continues to grow, the demand for developing more efficient social network-based recommendation approaches will continue to grow as well. In this paper, following proven optimization techniques from the domain of machine learning with constrained optimization, and modifying them accordingly in order to take into account the social network information, we propose a matrix factorization algorithm that improves on previously proposed related approaches in terms of convergence speed, recommendation accuracy and performance on cold start users. The proposed algorithm can be implemented easily, and thus used more frequently in social recommendation setups. Our claims are validated by experiments on two real life data sets, the public domain Epinions.com dataset and a much larger dataset crawled from Flixster.com. Full article

The application of Empirical Line Method (ELM) for hyperspectral Atmospheric Compensation (AC) premises the underlying linear relationship between a material’s reflectance and appearance. ELM solves the Radiative Transfer (RT) equation under specialized constraint by means of in-scene white and black calibration panels. The reflectance of material is invariant to illumination. Exploiting this property, we articulated a mathematical formulation based on the RT model to create cost functions relating variably illuminated regions within a scene. In this paper, we propose multi-layered regression learning-based recovery of radiance components, i.e., total ground-reflected radiance and path radiance from reflectance and radiance images of the scene. These decomposed components represent terms in the RT equation and enable us to relate variable illumination. Therefore, we assume that Hyperspectral Image (HSI) radiance of the scene is provided and AC can be processed on it, preferably with QUick Atmospheric Correction (QUAC) algorithm. QUAC is preferred because it does not account for surface models. The output from the proposed algorithm is an intermediate map of the scene on which our mathematically derived binary and multi-label threshold is applied to classify shadowed and non-shadowed regions. Results from a satellite and airborne NADIR imagery are shown in this paper. Ground truth (GT) is generated by ray-tracing on a LIDAR-based surface model in the form of contour data, of the scene. Comparison of our results with GT implies that our algorithm’s binary classification shadow maps outperform other existing shadow detection algorithms in true positive, which is the detection of shadows when it is in ground truth. It also has the lowest false negative i.e., detecting non-shadowed region as shadowed, compared to existing algorithms. Full article

Recognizing objects and estimating their poses have a wide range of application in robotics. For instance, to grasp objects, robots need the position and orientation of objects in 3D. The task becomes challenging in a cluttered environment with different types of objects. A popular approach to tackle this problem is to utilize a deep neural network for object recognition. However, deep learning-based object detection in cluttered environments requires a substantial amount of data. Collection of these data requires time and extensive human labor for manual labeling. In this study, our objective was the development and validation of a deep object recognition framework using a synthetic depth image dataset. We synthetically generated a depth image dataset of 22 objects randomly placed in a 0.5 m × 0.5 m × 0.1 m box, and automatically labeled all objects with an occlusion rate below 70%. Faster Region Convolutional Neural Network (R-CNN) architecture was adopted for training using a dataset of 800,000 synthetic depth images, and its performance was tested on a real-world depth image dataset consisting of 2000 samples. Deep object recognizer has 40.96% detection accuracy on the real depth images and 93.5% on the synthetic depth images. Training the deep learning model with noise-added synthetic images improves the recognition accuracy for real images to 46.3%. The object detection framework can be trained on synthetically generated depth data, and then employed for object recognition on the real depth data in a cluttered environment. Synthetic depth data-based deep object detection has the potential to substantially decrease the time and human effort required for the extensive data collection and labeling. Full article

In recent years researchers have worked to understand image contents in computer vision. In particular, the bag of visual words (BoVW) model, which describes images in terms of a frequency histogram of visual words, is the most adopted paradigm. The main drawback is the lack of information about location and the relationships between features. For this purpose, we propose a new paradigm called bag of ARSRG (attributed relational SIFT (scale-invariant feature transform) regions graph) words (BoAW). A digital image is described as a vector in terms of a frequency histogram of graphs. Adopting a set of steps, the images are mapped into a vector space passing through a graph transformation. BoAW is evaluated in an image classification context on standard datasets and its effectiveness is demonstrated through experimental results compared with well-known competitors. Full article

The need to detect outliers or otherwise unusual data, which can be formalized as the estimation a particular quantile of a distribution, is an important problem that frequently arises in a variety of applications of pattern recognition, computer vision and signal processing. For example, our work was most proximally motivated by the practical limitations and requirements of many semi-automatic surveillance analytics systems that detect abnormalities in closed-circuit television (CCTV) footage using statistical models of low-level motion features. In this paper, we specifically address the problem of estimating the running quantile of a data stream with non-stationary stochasticity when the absolute (rather than asymptotic) memory for storing observations is severely limited. We make several major contributions: (i) we derive an important theoretical result that shows that the change in the quantile of a stream is constrained regardless of the stochastic properties of data; (ii) we describe a set of high-level design goals for an effective estimation algorithm that emerge as a consequence of our theoretical findings; (iii) we introduce a novel algorithm that implements the aforementioned design goals by retaining a sample of data values in a manner adaptive to changes in the distribution of data and progressively narrowing down its focus in the periods of quasi-stationary stochasticity; and (iv) we present a comprehensive evaluation of the proposed algorithm and compare it with the existing methods in the literature on both synthetic datasets and three large “real-world” streams acquired in the course of operation of an existing commercial surveillance system. Our results and their detailed analysis convincingly and comprehensively demonstrate that the proposed method is highly successful and vastly outperforms the existing alternatives, especially when the target quantile is high-valued and the available buffer capacity severely limited. Full article

Document-level sentiment analysis is a challenging task given the large size of the text, which leads to an abundance of words and opinions, at times contradictory, in the same document. This analysis is particularly useful in analyzing press articles and blog posts about a particular product or company, and it requires a high concentration, especially when the topic being discussed is sensitive. Nevertheless, most existing models and techniques are designed to process short text from social networks and collaborative platforms. In this paper, we propose a combination of Convolutional Neural Networks (CNN) and Bidirectional Long Short-Term Memory (BiLSTM) models, with Doc2vec embedding, suitable for opinion analysis in long texts. The CNN-BiLSTM model is compared with CNN, LSTM, BiLSTM and CNN-LSTM models with Word2vec/Doc2vec embeddings. The Doc2vec with CNN-BiLSTM model was applied on French newspapers articles and outperformed the other models with 90.66% accuracy. Full article

Point estimation of class prevalences in the presence of dataset shift has been a popular research topic for more than two decades. Less attention has been paid to the construction of confidence and prediction intervals for estimates of class prevalences. One little considered question is whether or not it is necessary for practical purposes to distinguish confidence and prediction intervals. Another question so far not yet conclusively answered is whether or not the discriminatory power of the classifier or score at the basis of an estimation method matters for the accuracy of the estimates of the class prevalences. This paper presents a simulation study aimed at shedding some light on these and other related questions. Full article

Graphs are a very useful framework for representing information. In general, these data structures are used in different application domains where data of interest are described in terms of local and spatial relations. In this context, the aim is to propose an alternative graph-based image representation. An image is encoded by a Region Adjacency Graph (RAG), based on Multicolored Neighborhood (MCN) clustering. This representation is integrated into a Content-Based Image Retrieval (CBIR) system, designed for the vision-based positioning task. The image matching phase, in the CBIR system, is managed with an approach of attributed graph matching, named the extended-VF algorithm. Evaluated in a context of indoor localization, the proposed system reports remarkable performance. Full article

The analysis of knee kinematic data, which come in the form of a small sample of discrete curves that describe repeated measurements of the temporal variation of each of the knee three fundamental angles of rotation during a subject walking cycle, can inform knee pathology classification because, in general, different pathologies have different kinematic data patterns. However, high data dimensionality and the scarcity of reference data, which characterize this type of application, challenge classification and make it prone to error, a problem Duda and Hart refer to as the curse of dimensionality. The purpose of this study is to investigate a sample-based classifier which evaluates data proximity by the two-sample Hotelling $T^2$ statistic. This classifier uses the whole sample of an individual’s measurements for a better support to classification, and the Hotelling $T^2$ hypothesis testing made applicable by dimensionality reduction. This method was able to discriminate between femero-rotulian (FR) and femero-tibial (FT) knee osteoarthritis kinematic data with an accuracy of $88.1\%$ , outperforming significantly current state-of-the-art methods which addressed similar problems. Extended to the much harder three-class problem involving pathology categories FR and FT, as well as category FR-FT which represents the incidence of both diseases FR and FT in a same individual, the scheme was able to reach a performance that justifies its further use and investigation in this and other similar applications. Full article

This paper presents a semi-supervised faster region-based convolutional neural network (SF-RCNN) approach to detect persons and to classify the load carried by them in video data captured from distances several miles away via high-power lens video cameras. For detection, a set of computationally efficient image processing steps are considered to identify moving areas that may contain a person. These areas are then passed onto a faster RCNN classifier whose convolutional layers consist of ResNet50 transfer learning. Frame labels are obtained in a semi-supervised manner for the training of the faster RCNN classifier. For load classification, another convolutional neural network classifier whose convolutional layers consist of GoogleNet transfer learning is used to distinguish a person carrying a bundle from a person carrying a long arm. Despite the challenges associated with the video dataset examined in terms of the low resolution of persons, the presence of heat haze, and the shaking of the camera, it is shown that the developed approach outperforms the faster RCNN approach. Full article