Search

We address the problem of 3D protein deformable shape classification. Proteins are macromolecules characterized by deformable and complex shapes which are related to their function making their classification an important task. Their molecular surface is represented by graphs such as triangular tessellations or meshes. In this paper, we propose a new graph embedding based approach for the classification of these 3D deformable objects. Our technique is based on graphs decomposition into a set of substructures, using triangle-stars, which are subsequently matched with the Hungarian algorithm. The proposed approach is based on an approximation of the Graph Edit Distance which is characterized by its robustness against both noise and distortion. Our algorithm defines a metric space using graph embedding techniques, where each object is represented by a set of selected 3D prototypes. We propose new approaches for prototypes selection and features reduction. The classification is performed with supervised machine learning techniques. The proposed method is evaluated against 3D protein benchmark repositories and state-of-the-art algorithms. Our experimental results consistently demonstrate the effectiveness of our approach., 4th International Conference on Imaging, Vision & Pattern Recognition (IVPR-2020), June 22-25, 2020, Kitakyushu, Japan

In e-business, recommender systems have been instrumental in guiding users through their online experiences. However, these systems are often limited by the lack of labels data and data sparsity. Increasingly, data-mining techniques are utilized to address this issue. In most research, recommendations to be made are achieved via supervised learning that typically employs the k-nearest neighbor learner. However, supervised learning relies on labeled data, which may not be available at the time of learning. Data sparsity, which refers to situations where the number of items that have been recommended represents only a small subset of all available items, further affects model performance. One suggested solution is to apply cluster analysis as a preprocessing step and thus guide the learning process from natural grouping, typically using similar customer profiles, to improve predictive accuracy. In this paper, we study the benefits of applying cluster analysis as a preprocessing step prior to constructing classification models. Our HCC-Learn framework combines content-based analysis in the preprocessing stage and collaborative filtering in the final prediction stage. Our results show the value of our HCC-Learn framework applied to real-world data sets, especially when combining soft clustering and ensembles based on feature subspaces., 10th International Joint Conference, IC3K 2018, September 18-20, 2018, Seville, Spain, Series: Communications in Computer and Information Science

11th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2019), Sept. 17-19, 2019, Vienna, Austria, Series: Communications in Computer and Information Science

Pathologic complete response (pCR) is a predictor for favorable outcome after neoadjuvant treatment in early breast cancer. Modulation of gene expression may also provide early readouts of biological activity and prognosis, offering the possibility for timely response-guided treatment adjustment. The role of early transcriptional changes in predicting response to neoadjuvant chemotherapy plus bevacizumab was investigated. One-hundred-and-fifty patients with large, operable and locally advanced HER2-negative breast cancer received epirubicin and docetaxel, with the addition of bevacizumab. Patients underwent tumor biopsies at baseline, after Cycle 2 and at the time of surgery. The primary end point, pCR, and its relation with the secondary endpoints event-free survival (EFS), overall survival (OS) and gene expression profiles, are reported. The pCR rate was 13% (95% CI 8.6-20.2), with significantly more pCRs among triple-negative [28% (95% CI 14.8-45.4)] than among hormone receptor positive (HR+) tumors [9% (95% CI 4.6-16.3); (OR=3.9 [CI=1.5-10.3])]. pCR rates were not associated with EFS or OS. PAM50 subtypes significantly changed after Cycle 2 (p=0.03) and an index of absolute changes in PAM50 correlations between these time-points was associated with EFS [HR=0.62 (CI=0.3-1.1)]. In univariable analyses, signatures for angiogenesis, proliferation, estrogen receptor signaling, invasion and metastasis, and immune response, measured after Cycle 2, were associated with pCR in HR+ tumors. Evaluation of changes in molecular subtypes and other signatures early in the course of neoadjuvant treatment may be predictive of pCR and EFS. These factors may help guide further treatment and should be considered when designing neoadjuvant trials.

The process of continuously reallocating funds into financial assets, aiming to increase the expected return of investment and minimizing the risk, is known as portfolio management. In this paper, a portfolio management framework is developed based on a deep reinforcement learning framework called DeepBreath. The DeepBreath methodology combines a restricted stacked autoencoder and a convolutional neural network (CNN) into an integrated framework. The restricted stacked autoencoder is employed in order to conduct dimensionality reduction and features selection, thus ensuring that only the most informative abstract features are retained. The CNN is used to learn and enforce the investment policy which consists of reallocating the various assets in order to increase the expected return on investment. The framework consists of both offline and online learning strategies: the former is required to train the CNN while the latter handles concept drifts i.e. a change in the data distribution resulting from unforeseen circumstances. These are based on passive concept drift detection and online stochastic batching. Settlement risk may occur as a result of a delay in between the acquisition of an asset and its payment failing to deliver the terms of a contract. In order to tackle this challenging issue, a blockchain is employed. Finally, the performance of the DeepBreath framework is tested with four test sets over three distinct investment periods. The results show that the return of investment achieved by our approach outperforms current expert investment strategies while minimizing the market risk.

Currently, there is no marker in use in the clinical management of colon cancer to predict which patients will respond efficiently to 5‐fluorouracil (5‐FU), a common component of all cytotoxic therapies. Our aim was to develop and validate a multigene signature associated with clinical outcome from 5‐FU therapy and to determine if it could be used to identify patients who might respond better to alternate treatments. Using a panel of 5‐FU resistant and sensitive colon cancer cell lines, we identified 103 differentially expressed genes providing us with a 5‐FU response signature. We refined this signature using a clinically relevant DNA microarray‐based dataset of 359 formalin‐fixed and paraffin‐embedded (FFPE) colon cancer samples. We then validated the final signature in an external independent DNA microarray‐based dataset of 316 stage III FFPE samples from the PETACC‐3 (Pan‐European Trails in Alimentary Tract Cancers) clinical trial. Finally, using a drug sensitivity database of 658 cell lines, we generated a list of drugs that could sensitize 5‐FU resistant patients using our signature. We confirmed using the PETACC‐3 dataset that the overall survival of subjects responding well to 5‐FU did not improve with the addition of irinotecan (FOLFIRI; two‐sided log‐rank test p = 0.795). Conversely, patients who responded poorly to 5‐FU based on our 12‐gene signature were associated with better survival on FOLFIRI therapy (one‐sided log‐rank test p = 0.039). This new multigene signature is readily applicable to FFPE samples and provides a new tool to help manage treatment in stage III colon cancer. It also provides the first evidence that a subgroup of colon cancer patients can respond better to FOLFIRI than 5‐FU treatment alone.

The identification of changes in data distributions associated with data streams is critical in understanding the mechanics of data generating processes and ensuring that data models remain representative through time. To this end, concept drift detection methods often utilize statistical techniques that take numerical data as input. However, many applications produce data streams containing categorical attributes, where numerical statistical methods are not applicable. In this setting, common solutions use error monitoring, assuming that fluctuations in the error measures of a learning system correspond to concept drift. Context-based concept drift detection techniques for categorical streams, which observe changes in the actual data distribution, have received limited attention. Such context-based change detection is arguably more informative as it is data-driven and directly applicable in an unsupervised setting. This paper introduces a novel context-based algorithm for categorical data, namely FG-CDCStream. In this unsupervised method, multiple drift detection tracks are maintained and their votes are combined in order to determine whether a real change has occurred. In this way, change detections are rapid and accurate, while the number of false alarms remains low. Our experimental evaluation against synthetic data streams shows that FG-CDCStream outperforms the state-of-the art. Our analysis further indicates that FG-CDCStream produces highly accurate and representative post-change models., Series: Lecture Notes in Computer Science; no. 10558

The success of data stream mining techniques has allowed decision makers to analyze their data in multiple domains, ranging from monitoring network intrusion to financial markets analysis and online sales transactions exploration. Specifically, online ensembles that construct accurate models against drifting data streams have been developed. Recently, there has been a surge in interest in mobile (or so-called pocket) data stream mining, aiming to construct near real-time models for data stream mining applications that run on mobile devices. In such a setting, it follows that the computational resources are limited and that there is a need to adapt analytics to map the resource usage requirements. Consequently, the resultant models should not only be highly accurate, but they should also adapt swiftly to changes. In addition, the data mining techniques should be fast, scalable, and efficient in terms of resource allocation. It then becomes important to consider Return on Investment (ROI) issues such as storage requirements and memory utilization. This paper introduces the Adaptive Ensemble Size (AES) algorithm, an extension of the Online Bagging method, to address these issues. Our AES method dynamically adapts the sizes of ensembles, based on ROI usage patterns. We illustrate our approach by analyzing the performances against both synthetic and real-world data streams. The results, when comparing our AES algorithm with the state-of-the-art, indicate that we are able to obtain a high Return on Investment (ROI) and to swiftly adapt to change, without compromising on the predictive accuracy.

Background The ability to reliably identify the state (activated, repressed, or latent) of any molecular process in the tumor of a patient from an individual whole-genome gene expression profile obtained from microarray or RNA sequencing (RNA-seq) promises important clinical utility. Unfortunately, all previous bioinformatics tools are only applicable in large and diverse panels of patients, or are limited to a single specific pathway/process (e.g. proliferation). Methods Using a panel of 4510 whole-genome gene expression profiles from 10 different studies we built and selected models predicting the activation status of a compendium of 1733 different biological processes. Using a second independent validation dataset of 742 patients we validated the final list of 1773 models to be included in a de novo tool entitled absolute inference of patient signatures (AIPS). We also evaluated the prognostic significance of the 1773 individual models to predict outcome in all and in specific breast cancer subtypes. Results We described the development of the de novo tool entitled AIPS that can identify the activation status of a panel of 1733 different biological processes from an individual breast cancer microarray or RNA-seq profile without recourse to a broad cohort of patients. We demonstrated that AIPS is stable compared to previous tools, as the inferred pathway state is not affected by the composition of a dataset. We also showed that pathway states inferred by AIPS are in agreement with previous tools but use far fewer genes. We determined that several AIPS-defined pathways are prognostic across and within molecularly and clinically define subtypes (two-sided log-rank test false discovery rate (FDR)

The ability to reliably identify the state (activated, repressed, or latent) of any molecular process in the tumor of a patient from an individual whole-genome gene expression profile obtained from microarray or RNA sequencing (RNA-seq) promises important clinical utility. Unfortunately, all previous bioinformatics tools are only applicable in large and diverse panels of patients, or are limited to a single specific pathway/process (e.g. proliferation). Using a panel of 4510 whole-genome gene expression profiles from 10 different studies we built and selected models predicting the activation status of a compendium of 1733 different biological processes. Using a second independent validation dataset of 742 patients we validated the final list of 1773 models to be included in a de novo tool entitled absolute inference of patient signatures (AIPS). We also evaluated the prognostic significance of the 1773 individual models to predict outcome in all and in specific breast cancer subtypes. We described the development of the de novo tool entitled AIPS that can identify the activation status of a panel of 1733 different biological processes from an individual breast cancer microarray or RNA-seq profile without recourse to a broad cohort of patients. We demonstrated that AIPS is stable compared to previous tools, as the inferred pathway state is not affected by the composition of a dataset. We also showed that pathway states inferred by AIPS are in agreement with previous tools but use far fewer genes. We determined that several AIPS-defined pathways are prognostic across and within molecularly and clinically define subtypes (two-sided log-rank test false discovery rate (FDR)

Great advances in analytical technology coupled with accelerated new drug development and growing understanding of biological challenges, such as tumor heterogeneity, have required a change in the focus for biobanking. Most current banks contain samples of primary tumors, but linking molecular signatures to therapeutic questions requires serial biopsies in the setting of metastatic disease, next-generation of biobanking. Furthermore, an integration of multidimensional analysis of various molecular components, that is, RNA, DNA, methylome, microRNAome and post-translational modifications of the proteome, is necessary for a comprehensive view of a tumor's biology. While data using such biopsies are now regularly presented, the preanalytical variables in tissue procurement and processing in multicenter studies are seldom detailed and therefore are difficult to duplicate or standardize across sites and across studies. In the context of a biopsy-driven clinical trial, we generated a detailed protocol that includes morphological evaluation and isolation of high-quality nucleic acids from small needle core biopsies obtained from liver metastases. The protocol supports stable shipping of samples to a central laboratory, where biopsies are subsequently embedded in support media. Designated pathologists must evaluate all biopsies for tumor content and macrodissection can be performed if necessary to meet our criteria of >60% neoplastic cells and

Objective To characterize at high resolution the DNA methylation changes which occur in the genome of serous epithelial ovarian cancer (EOC) in association with tumor aggressiveness. Methods Methylated DNA immunoprecipitation in combination with CpG island-tiling arrays was used to compare the methylation profiles of five borderline, five grade 1/stage III/IV, five grade 3/stage I and five grade 3/stage III/IV serous EOC tumors, to those of five normal human ovarian tissue samples. Results We found widespread DNA hypermethylation that occurs even in low-malignant potential (borderline) tumors and which predominantly includes key developmental/homeobox genes. Contrary to DNA hypermethylation, significant DNA hypomethylation was observed only in grade 3 serous EOC tumors. The latter observation was further confirmed when comparing the DNA methylation profiles of primary cell cultures derived from matched tumor samples obtained prior to, and following chemotherapy treatment from two serous EOC patients with advanced disease. To our knowledge this is the first report that has shown the presence of massive DNA hypomethylation in advanced serous EOC, associated with tumor malignancy and disease progression. Conclusions Our data raise the concern that demethylating drugs that are currently being used in advanced EOC disease (representing the majority of serous EOC cases) might have adverse effects due to activation of oncogenes and prometastatic genes. Understanding the relative roles of hypomethylation and hypermethylation in cancer could have clear implications on the therapeutic use of agents targeting the DNA methylation machinery.

Data streams, where an instance is only seen once and where a limited amount of data can be buffered for processing at a later time, are omnipresent in today’s real-world applications. In this context, adaptive online ensembles that are able to learn incrementally have been developed. However, the issue of handling data that arrives asynchronously has not received enough attention. Often, the true class label arrives after with a time-lag, which is problematic for existing adaptive learning techniques. It is not realistic to require that all class labels be made available at training time. This issue is further complicated by the presence of late-arriving, slowly changing dimensions (i.e., late-arriving descriptive attributes). The aim of active learning is to construct accurate models when few labels are available. Thus, active learning has been proposed as a way to obtain such missing labels in a data stream classification setting. To this end, this paper introduces an active online ensemble (AOE) algorithm that extends online ensembles with an active learning component. Our experimental results demonstrate that our AOE algorithm builds accurate models against much smaller ensemble sizes, when compared to traditional ensemble learning algorithms. Further, our models are constructed against small, incremental data sets, thus reducing the number of examples that are required to build accurate ensembles., 8th International Conference on Knowledge Discovery and Information Retrieval, November 9-11, 2016, Porto, Portugal

Online ensemble methods have been very successful to create accurate models against data streams that are susceptible to concept drift. The success of data stream mining has allowed diverse users to analyse their data in multiple domains, ranging from monitoring stock markets to analysing network traffic and exploring ATM transactions. Increasingly, data stream mining applications are running on mobile devices, utilizing the variety of data generated by sensors and network technologies. Subsequently, there has been a surge in interest in mobile (or so-called pocket) data stream mining, aiming to construct near real-time models. However, it follows that the computational resources are limited and that there is a need to adapt analytics to map the resource usage requirements. In this context, the resultant models produced by such algorithms should thus not only be highly accurate and be able to swiftly adapt to changes. Rather, the data mining techniques should also be fast, scalable, and efficient in terms of resource allocation. It then becomes important to consider Return on Investment (ROI) issues such as storage space needs and memory utilization. This paper introduces the Adaptive Ensemble Size (AES) algorithm, an extension of the Online Bagging method, to address this issue. Our AES method dynamically adapts the sizes of ensembles, based on the most recent memory usage requirements. Our results when comparing our AES algorithm with the state-of-the-art indicate that we are able to obtain a high Return on Investment (ROI) without compromising on the accuracy of the results., 4th International Workshop on New Frontiers in Mining Complex Patterns, September 7, 2015, Porto, Portugal, Series: Lecture Notes in Computer Science; no. 9607

Data mining has been successfully applied in many businesses, thus aiding managers to make informed decisions that are based on facts, rather than having to rely on guesswork and incorrect extrapolations. Data mining algorithms equip institutions to predict the movements of financial indicators, enable companies to move towards more energy-efficient buildings, as well as allow businesses to conduct targeted marketing campaigns and forecast sales. Specific data mining success stories include customer loyalty prediction, economic forecasting, and fraud detection. The strength of data mining lies in the fact that it allows for not only predicting trends and behaviors, but also for the discovery of previously unknown patterns. However, a number of challenges remain, especially in this era of big data. These challenges are brought forward due to the sheer Volume of today’s databases, as well as the Velocity (in terms of speed of arrival) and the Variety, in terms of the various types of data collected. This chapter focuses on techniques that address these issues. Specifically, we turn our attention to the financial sector, which has become paramount to business. Our discussion centers on issues such as considering data distributions with high fluctuations, incorporating late arriving data, and handling the unknown. We review the current state-of-the-art, mainly focusing on model-based approaches. We conclude the chapter by providing our perspective as to what the future holds, in terms of building accurate models against today’s business, and specifically financial, data.

Computer-generated models of the human body generally do not adequately model the complex human morphology. These models, therefore, do not reflect the anthropometric realities and are not specific enough for commercial use. This paper presents an approach to adjust virtual models through the use of body measurements, as obtained from an anthropometric data set. In this approach, the measurements are used in grouping 3D body scans, which are obtained using precise optoelectronic measurement devices, into clusters. The virtual mannequins are then adjusted by using the measurements of the nearest cluster member. In this way, realistic accurate virtual mannequins are created.

We consider the problem of comparing deformable 3D objects represented by graphs, i.e., Triangular tessellations. We propose a new algorithm to measure the distance between triangular tessellations using a new decomposition of triangular tessellations into triangle-Stars. The proposed algorithm assures a minimum number of disjoint triangle-Stars, offers a better measure by covering a larger neighborhood and uses a set of descriptors which are invariant or at least oblivious under most common deformations. We prove that the proposed distance is a pseudo-metric. We analyse its time complexity and we present a set of experimental results which confirm the high performance and accuracy of our algorithm., 2015 International Conference on 3D Vision (3DV), October 19-22, 2015, Lyon, France

Aim: We determined whether the Y-box binding protein-1 (YB-1) and its binding partner, the X-linked ribosomal protein S4 (RPS4X), are associated with clinical outcome in bladder cancer. Materials & methods: A population of 167 patients with muscle-invasive bladder tumor without evidence of metastasis at time of cystectomy was analyzed retrospectively. YB-1 and RPS4X expressions were evaluated immunohistochemically in tumors and analyzed for association with clinical variables and survival. Results: Kaplan–Meier and multivariate Cox regression analyses indicated that low expression of RPS4X was associated with a higher risk of death or disease recurrence. In contrast, YB-1 was not significantly associated with either recurrence-free or overall survival. Conclusion: Low RPS4X expression is associated with poor disease-specific and recurrence-free survival in bladder cancer.

Class imbalance is a crucial problem in machine learning and occurs in many domains. Specifically, the two-class problem has received interest from researchers in recent years, leading to solutions for oil spill detection, tumour discovery and fraudulent credit card detection, amongst others. However, handling class imbalance in datasets that contains multiple classes, with varying degree of imbalance, has received limited attention. In such a multi-class imbalanced dataset, the classification model tends to favour the majority classes and incorrectly classify instances from the minority classes as belonging to the majority classes, leading to poor predictive accuracies. Further, there is a need to handle both the imbalances between classes as well as address the selection of examples within a class (i.e. the so-called within class imbalance). In this paper, we propose the SCUT hybrid sampling method, which is used to balance the number of training examples in such a multi-class setting. Our SCUT approach oversamples minority class examples through the generation of synthetic examples and employs cluster analysis in order to undersample majority classes. In addition, it handles both within-class and between-class imbalance. Our experimental results against a number of multi-class problems show that, when the SCUT method is used for pre-processing the data before classification, we obtain highly accurate models that compare favourably to the state-of-the-art., The 7th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management, November 12-14, 2015, Lisbon, Portugal

Non-rigid shapes are generally known as objects where the three dimensional geometry may deform by internal and/or external forces. Deformable shapes are all around us, ranging from macromolecules, to natural objects such as the trees in the forest or the fruits in our gardens, and even human bodies. The development of measurements to accurately describe such non-rigid shapes has wide application. It follows that, when aiming to perform measurement of non-rigid shapes, one need to ensure that the intrinsic geometries are preserved in the presence of deformations. This paper introduces physics-based techniques for the measurements of such deformable shapes. Our work is based on the idea of heat propagation through a surface. Specifically, we introduce so-called multi-measurements to measure the distance between two points on an object and reflective measurements that allows for the exploration of a region around a point. Further, we introduce two novel meta-measurements that extend the bag of features concept. We illustrate our methodology in the context of structural proteomics and molecular docking., 2014 IEEE International Conference on Computational Intelligence and Virtual Environments for Measurement Systems and Applications, CIVEMSA 2014, 5 May 2014 through 7 May 2014, Ottawa, ON

Summary: Breast carcinoma (BC) has been extensively profiled by high-throughput technologies for over a decade, and broadly speaking, these studies can be grouped into those that seek to identify patient subtypes (studies of heterogeneity) or those that seek to identify gene signatures with prognostic or predictive capacity. The sheer number of reported signatures has led to speculation that everything is prognostic in BC. Here, we show that this ubiquity is an apparition caused by a poor understanding of the interrelatedness between subtype and the molecular determinants of prognosis. Our approach constructively shows how to avoid confounding due to a patient’s subtype, clinicopathological profile, or treatment profile. The approach identifies patients who are predicted to have good outcome at time of diagnosis by all available clinical and molecular markers but who experience a distant metastasis within 5 years. These inherently difficult patients (∼7% of BC) are prioritized for investigations of intratumoral heterogeneity. : Tofigh et al. perform a comprehensive and systematic comparison of existing prognostic signatures in breast cancer. The comparison establishes that the supposed ubiquity of prognostic genes and processes has been grossly overestimated and results from insufficient stratification by subtype and other clinicopathological variables. The study presents a refined subtyping scheme ablating these effects. Nevertheless, a set of patients is identified whose outcome appears inherently difficult to predict using all available information at time of diagnosis.

The description of the spatial characteristics of two- and three-dimensional objects, in the framework of MPEG-7, is considered. The shape of an object is one of its fundamental properties, and this paper describes an efficient way to represent the coarse shape, scale and composition properties of an object. This representation is invariant to resolution, translation and rotation, and may be used for both two-dimensional (2-D) and three-dimensional (3-D) objects. This coarse shape descriptor will be included in the eXperimentation Model (XM) of MPEG-7. Applications of such a description to search object databases, in particular the CAESAR anthropometric database are discussed.

Cleopatra is a 3D database management software to classify the digitized 3D shape of the human bodies contained in the CAESAR database. The classification process is based on the geometric characteristics of the digitized data, along with a filtering process based on text information contained in both, the traditional anthropometric and the extracted measurements files. The information contained in the demographic file is also available for filtering. This paper discusses the experimental results of tests related to the subject's pose while the scan is made. This is a repeatability test where the subjects (a male and a female) are repeatedly scanned ten times. Each scan is made at different times and the subject attempts to assume the same pose each time.

New methods for recognition of range images based on correlation are presented. The techniques allow the detection of a range image of a tridimensional pattern whatever its position along the x, y and z axes. The first method uses the phase-coding of the range image in order to obtain invariance along the z-axis in the correlation process. The second method uses the sine-coding of the range image in order to obtain better discrimination than that obtained with phase-coding. The third method is a hybrid one that makes use of sine-coding for the input scene and phase-coding for the reference image, so both the full translation invariance and high discrimination abilities are obtained. Computer simulations are presented. Also, an experimental implementation of the method has been carried out showing good agreement with theoretical predictions. On presente de nouvelles methodes pour reconnaitre des images en distance basees sur la correlation. Ces techniques permettent la detection d'une image en distance d'une structure tridimensionnelle quelque soit sa position le long de axes x, y et z. La premiere methode utilise le codage de phase d'une telle image pour obtenir l'invariance selon l'axe z dans le processus de correlation. La seconde methode utilise un codage sinusoidal de l'image pour obtenir une meilleure discrimination qu'avec la methode precedente. La troisieme est une methode hybride qui utilise le codage sinusoidal pour la scene d'entee et le codage de phase par l'image de reference, ainsi on obtient a la fois l'invariance par translation et de fortes possibilites de discrimination. On montre des simulations numeriques, et aussi un montage experimental traduisant cette methode et indiquant un bon accord avec les previsions theoriques.

Algorithms allowing the classification of three-dimensional (3D) polychromatic images based on amplitude modulation and amplitude-shift keying are introduced. For this purpose, the 3D information is likened to a `carrier wave?, while the color information is likened to a `modulating signal?. The first algorithm is based on the amplitude modulation of the range by the hue. On the amplitude-shift keying platform, two algorithms are designed based on a three-channels red, green and blue modulation and a six-channels hue modulation. In the three-channels approach, the modulation of the three-dimensional image is performed on the basis of the relative ordination of the red, green and blue components, while in the six-channels approach, modulation is performed by partitioning the hue interval into six subintervals. The cross-angle distributions of the modulated range images are calculated in order to obtain invariant representations. Shift keying modulation is also applied to the cross-angle distributions in order to obtain a chromatic sampling of cross-angles. From these cross-angles representations a set of histograms are built. The classification is performed by comparing the unknown histograms to a reference set. The classified object corresponds to the smallest error.

Acute myeloid leukemia (AML) with inversion of chromosome 3 is characterized by overexpression of EVI1 and carries a dismal prognosis. Arsenic-containing compounds have been described to be efficacious in malignancies overexpressing EVI1. Here, we describe a case of AML with inv(3)(q21q26.2) treated with the organic arsenical darinaparsin. Using a “personalized medicine approach,” two different arsenicals were screened for anti-leukemic effect against the patient’s cells ex vivo. The most promising compound, darinaparsin, was selected for in vivo treatment. Clinical effect was almost immediate, with a normalization of temperature, a stabilization of white blood cell (WBC) counts and an increased quality of life. Longitudinal monitoring of patient response and resistance incorporating significant correlative studies on patient-derived blood samples over the two cycles of darinaparsin given to this patient allowed us to evaluate potential mechanisms of response and resistance. The anti-leukemic effects of darinaparsin correlated with inhibition of the alternative NF-κB pathway and production of the inflammatory cytokine IL-8. Emergence of resistance was suspected during treatment cycle 2 and supported by xenograft studies in nude mice. Darinaparsin resistance correlated with an attenuation of the effect of treatment on the alternative NF-κB pathway. The results from this patient indicate that darinaparsin may be a good treatment option for inv(3) AML and that inhibition of the alternative NF-κB pathway may be predictive of response. Longitudinal monitoring of disease response as well as several correlative parameters allowed for the generation of novel correlations and predictors of response to experimental therapy in a heavily pretreated patient.

Consider a multi-relational database, to be used for classification, that contains a large number of unlabeled data. It follows that the cost of labeling such data is prohibitive. Transductive learning, which learns from labeled as well as from unlabeled data already known at learning time, is highly suited to address this scenario. In this paper, we construct multi-views from a relational database, by considering different subsets of the tables as contained in a multi-relational database. These views are used to boost the classification of examples in a co-training schema. The automatically generated views allow us to overcome the independence problem that negatively affect the performance of co-training methods. Our experimental evaluation empirically shows that co-training is beneficial in the transductive learning setting when mining multi-relational data and that our approach works well with only a small amount of labeled data., 8th International Conference on Machine Learning and Data Mining (MLDM) 2012, July 13-20, 2012 Berlin, Germany, Series: Lecture Notes in Computer Science; no. 7376

The apparel industry aims to produce comfortable and aesthetically pleasing garments that fit populations well. However, repeated studies of apparel customers' levels of satisfaction indicate that their needs are often not being met. In order to produce better fitting clothes, it is crucial to understand the body profiles of typical consumers. Exploring the demographic profiles of correctly identified customer segments holds obvious benefit for production, marketing and ensuring return on investments. To this end, we explore a database, containing anthropometric measurements, demographic profiles and 3-D body scans, of samples of the North American, Italian and Dutch populations. Through the use of objective interestingness measures-based feature selection and feature extraction, we accurately discover the relevant subsets of body measurements that require special care when designing clothes. Furthermore, we apply association analysis to the demographic data, in order to target customers and thus potentially defining new market opportunities. Copyright © 2012 Inderscience Enterprises Ltd.

Werner syndrome is a premature aging disorder caused by mutations in a RecQ-like DNA helicase. Mice lacking the helicase domain of the WRN homologue exhibit many features of Werner syndrome, including a pro-oxidant status and a shorter mean life span. Here, we show that resveratrol supplementation improved the hyperglycemia and the insulin resistance phenotype in these Wrn mutant mice. In addition, resveratrol reversed liver steatosis, lipid peroxidaton, and the defenestration phenotypes observed in such mice. Resveratrol, however, did not improve the hypertriglyceridemia, inflammatory stress, nor extend the mean life span of these mutant mice. Microarray and biologic pathway enrichment analyses on liver tissues revealed that resveratrol mainly decreased lipidogenesis and increased genes involved in the insulin signaling pathway and the glutathione metabolism in Wrn mutant mice. Finally, resveratrol-treated mutant mice exhibited an increase in the frequency of lymphoma and of several solid tumors. These results indicate that resveratrol supplementation might exert at least metabolic benefits for Werner syndrome patients.

Clothes should fit consumers well, be aesthetically pleasing and comfortable. However, repeated studies of customers' levels of satisfaction indicate that this is often not the case. For example, more robust males often find it difficult to find pants that are the correct length and fit their waists well. What, then, are the typical body profiles of the population? Would it be possible to identify the measurements that are of importance for different sizes and genders? Furthermore, assuming that we have access to an anthropometric database would there be a way to guide the data mining process to discover only those relevant body measurements that are of the most interest for apparel designers? This paper describes our results when addressing these questions through cluster analysis and interestingness measures-based feature selection. We explore a database containing anthropometric measurements as well as 3-D body scans, of a representative sample of the Dutch population.

Long-term digital preservation, the process of maintaining digital objects through time to ensure continued access, has become a crucial issue in recent years. Whilst the amount of digitised information is constantly increasing, so too is the pace of progress in information technology, resulting in obsolescence of the software and hardware required to access and view digital information. Despite many organisations recognising this threat and the resulting need for preservation action, more work is required to effectively address the issue. We present in this article a framework for the long-term digital preservation of 3-D data. This framework is based on two pertinent preservation practices, emulation and metadata which ensure that the authenticity and usability, respectively, of a preserved digital object remain intact through time. An evaluation of our framework is presented which illustrates the viability of our approach in retaining accessibility, authenticity and usability for future end users., Mr. Paquet was affiliated with the National Research Council of Canada, Visual Information Technology when he authored this article.

The current explosion of data and information, which are mainly caused by the continuous adoption of data warehouses and the extensive use of the Internet and its related technologies, has increased the urgent need for the development of techniques for intelligent data analysis. Data mining, which concerns the discovery and extraction of knowledge chunks from large data repositories, addresses this need. Data mining automates the discovery of hidden patterns and relationships that may not always be obvious. Data mining tools include classification techniques (such as decision trees, rule induction programs and neural networks) (Kou et al., 2007); clustering algorithms and association rule approaches, amongst others. Data mining has been fruitfully used in many of domains, including marketing, medicine, finance, engineering and bioinformatics. There still are, however, a number of factors that militate against the widespread adoption and use of this new technology. This is mainly due to the fact that the results of many data mining techniques are often difficult to understand. For example, the results of a data mining effort producing 300 pages of rules will be difficult to analyze. The visual representation of the knowledge embedded in such rules will help to heighten the comprehensibility of the results. The visualization of the data itself, as well as the data mining process should go a long way towards increasing the user’s understanding of and faith in the data mining process. That is, data and information visualization provide users with the ability to obtain new insights into the knowledge, as discovered from large repositories. This paper describes a number of important visual data mining issues and introduces techniques employed to improve the understandability of the results of data mining. Firstly, the visualization of data prior to, and during, data mining is addressed. Through data visualization, the quality of the data can be assessed throughout the knowledge discovery process, which includes data preprocessing, data mining and reporting. We also discuss information visualization, i.e. how the knowledge, as discovered by a data mining tool, may be visualized throughout the data mining process. This aspect includes visualization of the results of data mining as well as the learning process. In addition, the paper shows how virtual reality and collaborative virtual environments may be used to obtain an immersive perspective of the data and the data mining process as well as how visual data mining can be used to directly mine functionality with specific applications in the emerging field of proteomics.

The issue of long term preservation of digital data has, in recent years, become a critical issue with many diverse groups and organisations recognising the need to preserve digital documents before they fall victim to digital obsolescence. However, there is a distinct lack of discussion regarding the needs of future end users of such preserved documents. This is surprising, given that it is the end user who ultimately determines what information should be archived and in what way. Furthermore, it is essential that a preserved document retains its authenticity and usability through time and is easily interpreted by future end users. To this extent, we present in this paper details of our case study: an emulation framework to preserve 2D and 3D anthropometric data. We describe the testing of this emulation environment, by anthropometric experts, against the original environment and compare the two based on a given set of usability and authenticity criteria.

Multi-relational data mining methods discover patterns across multiple interlinked tables (relations) in a relational database. In many large organizations, such a multi-relational database spans numerous departments and/or subdivisions, which are involved in different aspects of the enterprise such as customer profiling, fraud detection, inventory management, financial management, and so on. When considering multi-relational classification, it follows that these subdivisions will express different interests in the data, leading to the need to explore various subsets of relevant relations with high utility with respect to the target class. The paper presents a novel approach for pruning the uninteresting relations of a relational database where relations come from such different parties and spans many classification tasks. We aim to create a pruned structure and thus minimize predictive performance loss on the final classification model. Our method identifies a set of strongly uncorrelated subgraphs to use for training and discards all others. The experiments performed demonstrate that our strategy is able to significantly reduce the size of the relational schema without sacrificing predictive accuracy., The 11th European Conference on Principles and Practice of Knowledge Discovery in Databases (PKDD), September 17-21, 2007, Warsaw, Poland

Clothes should be designed to tailor well, fit the body elegantly and hide obvious body flaws. To attain this goal, it is crucial to know the interrelationships between different body measurements, such as the interplay between e.g. shoulder width, neck circumference and waist. This paper discusses a study to better understand the typical consumer, from a virtual tailor's perspective. Cluster analysis was used to group the population into five clothing sizes. Next, multi-relational classification was applied to analyze the interplay between each group's anthropometric body measurements. Throughout this study, three-dimensional (3-D) body scans were used to verify the validity of our findings. Our results indicate that different sets of body measurements are used to characterize each clothing size. This information, together with the demographic profiles of the typical consumer, provides us with new insight into our evolving population., The 17th International European Conference on Machine Learning and the 10th European Conference on Principles and Practice of Knowledge Discovery in Databases (ECML/PKDD 2006), September 18-22, 2006, Berlin, Germany

Computer generated models of the human body generally do not adequately model the complex human morphology. These models therefore do not reflect the anthropometric realities and are not specific enough for commercial use. This paper presents an approach to calibrate virtual models through the use of body measurements as obtained from an anthropometric data set. In this approach, the measurements are used to group three-dimensional body scans into clusters. The virtual mannequins are then calibrated by using the measurements of nearest cluster member. In this way, highly realistic, accurate virtual mannequins are created

The National Research Council of Canada (NRC) has developed a range of 3D imaging technology tools, which have been applied to a wide range of museum and heritage recording applications. The technology suite includes the development of high-resolution laser scanner systems as well as software for the preparation of accurate 3D models and for the display, analysis and comparison of 3D data. This paper will offer an overview of the technology and its museum and heritage applications with particular reference to the 3D examination of paintings and recording of archaeological sites.

The current explosion of data and information, mainly caused by data warehousing technologies as well as the extensive use of the Internet and its related technologies, has increased the urgent need for the development of techniques for intelligent data analysis. Data mining, which concerns the discovery and extraction of knowledge chunks from large data repositories, is aimed at addressing this need. Data mining automates the discovery of hidden patterns and relationships that may not always be obvious. Data mining tools include classification techniques (such as decision trees, rule induction programs and neural networks) (Han & Kamber, 2001), clustering algorithms and association rule approaches, amongst others.

The CAESAR study (Civilian American and European Surface Anthropometry Resource) is a survey of body measurements for people ages 18-65 in three countries: USA, The Netherlands, and Italy. It is a collaborative effort with partners from several countries and many industries. It employs the latest in 3D measurement technology, and is the first 3D surface anthropometry survey of the US and Europe. Data from the study will allow us to effectively design everything from better fitting clothes and protective equipment to better seats and workstations, as well as dramatically reducing the cost to design and manufacture them.

Content based management of three dimensional data in the framework of the MPEG-7 standard is considered. An overview of MPEG-7 and three dimensional data is presented. Descriptors and description schemes suitable for three dimensional data are introduced and their integration in the framework of MPEG-7 is detailed. The shape descriptors are based on cords, wavelet transform and three dimensional statistical moments. Implications of the normative and the non-normative parts of the standard are analyzed and their relations are considered. The descriptors and description schemes are evaluated according to MPEG-7 criteria. Experimental results indicating the potential of the descriptors are presented. Some considerations on dynamic scenes are introduced.

This chapter presents an approach for multimedia information retrieval employing an image-based algorithm. The Virtual Boutique is a novel, content-based system for multimedia information retrieval, based on virtual reality and visual information management. It supports realistic and fast rendering, extensive manipulation, reconfiguration and multimedia searching capabilities. The Virtual Boutique includes algorithms for searching through the inventory of an e-business, utilizing both images and three-dimensional objects. The chapter shows the application of The Virtual Boutique to an online boutique. In addition, it is illustrated how data mining can be used to determine patterns in consumer-interaction

Data mining concerns the discovery and extraction of knowledge chunks from large data repositories. In a cooperative datamining environment, more than one data mining tool collaborates during the knowledge discovery process. This chapter describes a data mining approach used to visualize the cooperative data mining process. According to this approach, visual data mining consists of both data and knowledge visualization. First, the data are visualized during both data preprocessing and data mining. In this way, the quality of the data is assessed and improved throughout the knowledge discovery process. Second, the knowledge, as discovered by the individual learners, is assessed and modified through the interactive visualization of the cooperative data mining process and its results. The knowledge obtained from the human domain expert also forms part of the process. Finally, the use of virtual reality-based visualization is proposed as a new method to model both the data and its descriptors.

Last year we presented the influence of pose on threedimensional (3-D) shape classification in the context of a repeatability study. Meaning, that the subjects are repeatedly scanned 10 times and they attempt to assume the same pose each time. It was shown that changes due to a slight pose modification had no detrimental effects for shape classification. This paper discuss a second set of experiments, designed to test the stability of the geometric search engine in more extreme cases. The pose of the subjects between the scans are modified substantially, with an increasing amount of differences compared to the CAESAR protocol. Experimental results will be presented and discussed., Proceedings of the Digital Human Modeling for Design and Engineering Conference, June 26-28, 2001., Arlington, Virginia, USA

The Virtual Boutique is made out of three modules: the decor, the market and the search engine. The decor is the physical space occupied by the Virtual Boutique. It can reproduce any existing boutique. For this purpose, photogrammetry is used. A set of pictures of a real boutique or space is taken and a virtual 3D representation of this space is calculated from them. Calculations are performed with software developed at NRC. This representation consists of meshes and texture maps. The camera used in the acquisition process determines the resolution of the texture maps. Decorative elements are added like painting, computer generated objects and scanned objects. The objects are scanned with laser scanner developed at NRC. This scanner allows simultaneous acquisition of range and color information based on white laser beam triangulation. The second module, the market, is made out of all the merchandises and the manipulators, which are used to manipulate and compare the objects. The third module, the search engine, can search the inventory based on an object shown by the customer in order to retrieve similar objects base don shape and color. The items of interest are displayed in the boutique by reconfiguring the market space, which mean that the boutique can be continuously customized according to the customer's needs. The Virtual Boutique is entirely written in Java 3D and can run in mono and stereo mode and has been optimized in order to allow high quality rendering.

Proceedings of the Digital Human Modeling for Design Engineering, June 6-8, 2000., Dearborn, Michigan, USA

This paper presents a new approach for the classification and retrieval of three-dimensional images and models from databases. A set of retrieval algorithms is introduced. These algorithms are content-based, meaning that the input is not made out of keywords but of three-dimensional models. Tensor of inertia, distribution of normals, distribution of cords and multiresolution analysis are used to describe each model. The database can be searched by scale, shape or color or any combination of these parameters. A user friendly interface makes the retrieval operation simple and intuitive and allows to edit reference models according to the specifications of the user. Experimental results using a database of more than 400 range images and 1000 VRML models are presented.

Correlation methods for polychromatic range image recognition are presented. First, the range and the color are compared and the use of the hue images is discussed. The advantages and disadvantages in combining the range and the color information are analyzed. Single-channel and double-channel approaches are presented. The ability of both recognition methods to be invariant under translation in the xy plane and along the z axis are demonstrated. The codification of the range and the hue images using phase-coding, sine-coding and hybrid coding make possible the fully translation invariance. The double-channel is introduced using sine-coding images in the correlation process. Finally, the combination of those two channels is performed in order to improve the discrimination capability of the system. Digital results are shown.