Partition Tree Research Articles

Evolutionary Action Score Identifies a Subset of TP53 Mutated Myelodysplastic Syndrome with Favorable Prognosis

Evolutionary Action Score Identifies a Subset of TP53 Mutated Myelodysplastic Syndrome with Favorable Prognosis

Network Trees: A Method for Recursively Partitioning Covariance Structures.

In many areas of psychology, correlation-based network approaches (i.e., psychometric networks) have become a popular tool. In this paper, we propose an approach that recursively splits the sample based on covariates in order to detect significant differences in the structure of the covariance or correlation matrix. Psychometric networks or other correlation-based models (e.g., factor models) can be subsequently estimated from the resultant splits. We adapt model-based recursive partitioning and conditional inference tree approaches for finding covariate splits in a recursive manner. The empirical power of these approaches is studied in several simulation conditions. Examples are given using real-life data from personality and clinical research.

Groundwater contribution to transpiration of trees under wet and dry soil conditions*

AbstractThe objective of this study was to partition tree T under shallow groundwater table conditions into groundwater (Tgw) and soil water (Tsw). Three indigenous trees growing in lysimeters were installed with the compensation heat pulse velocity method to measure stem sap flow half‐hourly. Soil water content was monitored using HydraSCOUT probes. A groundwater table of 20 cm depth was maintained, and daily water replenishment to maintain a constant groundwater table was monitored using a graduated automatic supply bucket system. The contribution of Tgw towards total T for the investigated trees ranged from 31% when the topsoil was wet to 97% when it was dry. The contribution of Tsw ranged from 3% when the topsoil was dry to 69% after an irrigation event. It was concluded that trees switch to source water from superficial soil layers when the topsoil is wet and from groundwater during dry conditions.

Adaptive Feature Calculation and Diagonal Mapping for Successive Recovery of Tampered Regions

This article proposes an adaptive scheme for image tampered region localization and content recovery. To generate the watermark information comprised of the authentication data and recovery data, we firstly propose the Adaptive Authentication Feature Calculation algorithm to obtain the authentication data, which includes the information of block location and block feature. The DWT-based Block Feature Calculation method is then proposed to calculate the block feature, and the quantization method is employed to calculate the block location. The recovery data is composed of self-recovery bits and mapped-recovery bits. The self-recovery bits are obtained by the Set Partitioning in Hierarchical Trees encoding algorithm. For retrieving the damaged codes caused by tampering, we propose the Diagonal Mapping algorithm and apply it to the self-recovery bits, thus generating the mapped-recovery bits, to provide a guarantee of recovery data. Experimental results show the superior performance of the proposed scheme in terms of tamper detection and image recovery, by comparing with the state-of-the-art works. The results demonstrate that the proposed method shows efficiency in the adaptiveness, well localization, strong capability for image recovery, and the effectiveness of attack resistance.

Data-Driven Classification, Reduction, Parameter Identification and State Extension in Hybrid Power Systems

The paper describes a manifold learning-based algorithm for big data classification and reduction, as well as parameter identification in real-time operation of a power system. Both black-box and gray-box settings for SCADA- and PMU-based measurements are examined. Data classification is based on diffusion maps, where an improved data-informed metric construction for partition trees is used. Data classification and reduction is demonstrated on the measurement tensor example of calculated transient dynamics between two SCADA refreshing scans. Interpolation/extension schemes for state extension of restriction (from data to reduced space) and lifting (from reduced to data space) operators are proposed. The method is illustrated on the single-phase Motor D example from a very detailed WECC load model, connected to the single bus of a real-world 441-bus power system.

Complex study on compression of ECG signals using novel single-cycle fractal-based algorithm and SPIHT

Compression of ECG signal is essential especially in the area of signal transmission in telemedicine. There exist many compression algorithms which are described in various details, tested on various datasets and their performance is expressed by different ways. There is a lack of standardization in this area. This study points out these drawbacks and presents new compression algorithm which is properly described, tested and objectively compared with other authors. This study serves as an example how the standardization should look like. Single-cycle fractal-based (SCyF) compression algorithm is introduced and tested on 4 different databases—CSE database, MIT-BIH arrhythmia database, High-frequency signal and Brno University of Technology ECG quality database (BUT QDB). SCyF algorithm is always compared with well-known algorithm based on wavelet transform and set partitioning in hierarchical trees in terms of efficiency (2 methods) and quality/distortion of the signal after compression (12 methods). Detail analysis of the results is provided. The results of SCyF compression algorithm reach up to avL = 0.4460 bps and PRDN = 2.8236%.

Supervised quality evaluation of binary partition trees for object segmentation

The binary partition tree (BPT) allows for the hierarchical representation of images in a multiscale way, by providing a tree of nodes corresponding to image regions. In particular, cuts of a BPT can be interpreted as segmentations of the associated image. Building the BPT of an image then constitutes a relevant preliminary step for optimization-based segmentation methods. A wide literature has been devoted to the construction of BPTs, and their involvement in such segmentation tasks. Comparatively, there exist few works dedicated to evaluate the quality of BPTs, i.e. their ability to allow further segmentation methods to compute good results. We propose such a framework for evaluating the quality of a BPT with respect to the object segmentation problem, i.e. the segmentation of one or several objects from an image. This framework is supervised, since the notion of segmentation quality is not only depending on the application but also on the user’s objectives, expressed via the chosen ground-truth and quality metric. We develop two sides within this framework. First, we propose an intrinsic quality analysis, that relies on the structural coherence of the BPT with respect to ground-truth. More precisely, we evaluate to what extent the BPT structure is well-matching such examples, in a set / combinatorial fashion. Second, we propose an extrinsic analysis, by allowing the user to assess the quality of a BPT based on chosen metrics that correspond to the desired properties of the subsequent segmentation. In particular, we evaluate to what extent a BPT can provide good results with respect to such metrics whereas handling the trade-off with the cardinality of the cuts.

Native and exotic plant species respond differently to ecosystem characteristics at both local and landscape scales

Quantitative integration of factors that potentially affect exotic species richness and abundance at multiple spatial scales is relatively scarce in the literature. Our aim was to address this gap by evaluating the relative importance of the biotic community, abiotic factors, and landscape characteristics on the establishment and spread of native and exotic plant species. We assessed the effect of these factors on exotic and native species richness and abundance, and used regression tree and variation partitioning analyses to evaluate how these predictors interact to favor or limit exotic and/or native species. We found that landscape filters were especially important for the arrival of both native and exotic species, whereas biotic factors seemed to regulate the abundance of plant species once they were present within the system. However, the combined effects of different types of predictors explained the largest fraction of total variation in all models regarding exotic species. Furthermore, significant predictor variables had opposite effects on native versus exotic species at both local and landscape scales, which suggests that some ecosystem properties affect native and exotic species differently. Exotic species richness and abundance were increased by low values of native species cover and diversity, high landscape heterogeneity and edge density, human disturbances (e.g., mowing and soil disruption), land use activities (e.g., developed and agricultural areas), and proximity to transportation systems, especially highways. However, exotic species were less common in areas with low anthropogenic disturbance, where natural disturbances seemed to favor native plant species.

Biometric recognition system performance measures for lossy compression on EEG signals

Abstract Electroencephalogram (EEG) plays an essential role in analysing and recognizing brain-related diseases. EEG has been increasingly used as a new type of biometrics in person identification and verification systems. These EEG-based systems are important components in applications for both police and civilian works, and both areas process a huge amount of EEG data. Storing and transmitting these huge amounts of data are significant challenges for data compression techniques. Lossy compression is used for EEG data as it provides a higher compression ratio (CR) than lossless compression techniques. However, lossy compression can negatively influence the performance of EEG-based person identification and verification systems via the loss of information in the reconstructed data. To address this, we propose introducing performance measures as additional features in evaluating lossy compression techniques for EEG data. Our research explores if a common value of CR exists for different systems using datasets with lossy compression that could provide almost the same system performance with those using datasets without lossy compression. We performed experiments on EEG-based person identification and verification systems using two large EEG datasets, CHB MIT Scalp and Alcoholism, to investigate the relationship between standard lossy compression measures and our proposed system performance measures with the two lossy compression techniques, discrete wavelet transform—adaptive arithmetic coding and discrete wavelet transform—set partitioning in hierarchical trees. Our experimental results showed a common value of CR exists for different systems, specifically, 70 for person identification systems and 50 for person verification systems.

Machine Learning-Based Predictive Modeling of Postpartum Depression.

Postpartum depression is a serious health issue beyond the mental health problems that affect mothers after childbirth. There are no predictive tools available to screen postpartum depression that also allow early interventions. We aimed to develop predictive models for postpartum depression using machine learning (ML) approaches. We performed a retrospective cohort study using data from the Pregnancy Risk Assessment Monitoring System 2012–2013 with 28,755 records (3339 postpartum depression and 25,416 normal cases). The imbalance between the two groups was addressed by a balanced resampling using both random down-sampling and the synthetic minority over-sampling technique. Nine different ML algorithms, including random forest (RF), stochastic gradient boosting, support vector machines (SVM), recursive partitioning and regression trees, naïve Bayes, k-nearest neighbor (kNN), logistic regression, and neural network, were employed with 10-fold cross-validation to evaluate the models. The overall classification accuracies of the nine models ranged from 0.650 (kNN) to 0.791 (RF). The RF method achieved the highest area under the receiver-operating-characteristic curve (AUC) value of 0.884, followed by SVM, which achieved the second-best performance with an AUC value of 0.864. Predictive modeling developed using ML-approaches may thus be used as a prediction (screening) tool for postpartum depression in future studies.

A differential evolution based feature combination selection algorithm for high-dimensional data

Feature combination selection is used in object classification to select complementary features that can produce a powerful combination. One active area of selecting feature combinations is genome-wide association studies (GWAS). However, selecting feature combinations from high-dimensional GWAS data faces a serious issue of high computational complexity. In this paper, a fast evolutionary optimization method named search-history-guided differential evolution (HGDE) is proposed to deal with the problem. This method applies the search history memorized in a binary space partitioning tree to enhance its power for selecting feature combinations. We perform a comparative study on the proposed HGDE algorithm and other state-of-the-art algorithms using synthetic datasets, and later employ the HGDE algorithm in experiments on a real age-related macular degeneration dataset. The experimental results show that this proposed algorithm has superior performance in the selection of feature combinations. Moreover, the results provide a reference for studying the functional mechanisms of age-related macular degeneration.

Algorithm and VLSI Architecture Designs of a Lossless Embedded Compression Encoder for HD Video Coding Systems

The demand for visual quality has been advanced by high display resolutions and frame rates. Nevertheless, these two issues have caused tremendous memory bandwidth in a video coding system. In this study, an efficient lossless embedded compression (EC) algorithm is proposed to save memory bandwidth, while keeping visual quality. The proposed lossless EC algorithm incorporates three core techniques: tree partition, half-pixel prediction and group-based binary coding. Tree partition classifies a [Formula: see text] block into Trunk, Branch and Leaf. With tree partition, half-pixel prediction produces individual residues for Trunk, Branch and Leaf. Group-based binary coding converts theses residues to efficient codewords. The lossless compression ratio (CR) of the proposed EC is as high as 2.24 on average, saving memory bandwidth by 55.4%. This EC algorithm is implemented using CMOS 0.18[Formula: see text][Formula: see text]m technology. The maximum throughput can reach 6.4[Formula: see text]Gpixels/s, which can accommodate 3840 × 2160@60fps. The experiment results demonstrate that this study presents better hardware efficiency of 337[Formula: see text]Gpixels/J and 83.5[Formula: see text]Kpixels/s/gate.

Online Anomaly Detection With Bandwidth Optimized Hierarchical Kernel Density Estimators.

We propose a novel unsupervised anomaly detection algorithm that can work for sequential data from any complex distribution in a truly online framework with mathematically proven strong performance guarantees. First, a partitioning tree is constructed to generate a doubly exponentially large hierarchical class of observation space partitions, and every partition region trains an online kernel density estimator (KDE) with its own unique dynamical bandwidth. At each time, the proposed algorithm optimally combines the class estimators to sequentially produce the final density estimation. We mathematically prove that the proposed algorithm learns the optimal partition with kernel bandwidths that are optimized in both region-specific and time-varying manner. The estimated density is then compared with a data-adaptive threshold to detect anomalies. Overall, the computational complexity is only linear in both the tree depth and data length. In our experiments, we observe significant improvements in anomaly detection accuracy compared with the state-of-the-art techniques.

BROCELIANDE: A COMPARATIVE STUDY OF ATTRIBUTE PROFILES AND FEATURE PROFILES FROM DIFFERENT ATTRIBUTES

Abstract. Morphological attribute profiles (APs) are among the most effective spatial-spectral methods to perform multilevel image description based on hierarchical tree-based representation. They have been widely applied to the processing and characterization of remote sensing images, in particular to tackle classification task, in the literature. Recently, a novel extension of APs called FPs has been proposed by replacing pixel gray-levels with some statistical and geometrical features when forming the output profiles. FPs have been proved to be more efficient than the standard APs when generated from both inclusion and partition trees. The motivation of this article is to conduct a comparative study of APs and FPs using different attributes including some novel ones that have not been used in the literature. We also present our developed library called Broceliande, which proposes efficient implementation of APs and FPs to perform remote sensing image classification, with various choices of tree structures as well as attributes. We perform our experiments on two high resolution optical image data sets and provide comparative results of APs and FPs, showing and confirming their effectiveness to describe and classify remote sensing images.

Ensemble Modelling of Skipjack Tuna (Katsuwonus pelamis) Habitats in the Western North Pacific Using Satellite Remotely Sensed Data; a Comparative Analysis Using Machine-Learning Models

To examine skipjack tuna’s habitat utilization in the western North Pacific (WNP) we used an ensemble modelling approach, which applied a fisher- derived presence-only dataset and three satellite remote-sensing predictor variables. The skipjack tuna data were compiled from daily point fishing data into monthly composites and re-gridded into a quarter degree resolution to match the environmental predictor variables, the sea surface temperature (SST), sea surface chlorophyll-a (SSC) and sea surface height anomalies (SSHA), which were also processed at quarter degree spatial resolution. Using the sdm package operated in RStudio software, we constructed habitat models over a 9-month period, from March to November 2004, using 17 algorithms, with a 70:30 split of training and test data, with bootstrapping and 10 runs as parameter settings for our models. Model performance evaluation was conducted using the area under the curve (AUC) of the receiver operating characteristic (ROC), the point biserial correlation coefficient (COR), the true skill statistic (TSS) and Cohen’s kappa (k) metrics. We analyzed the response curves for each predictor variable per algorithm, the variable importance information and the ROC plots. Ensemble predictions of habitats were weighted with the TSS metric. Model performance varied across various algorithms, with the Support Vector Machines (SVM), Boosted Regression Trees (BRT), Random Forests (RF), Multivariate Adaptive Regression Splines (MARS), Generalized Additive Models (GAM), Classification and Regression Trees (CART), Multi-Layer Perceptron (MLP), Recursive Partitioning and Regression Trees (RPART), and Maximum Entropy (MAXENT), showing consistently high performance than other algorithms, while the Flexible Discriminant Analysis (FDA), Mixture Discriminant Analysis (MDA), Bioclim (BIOC), Domain (DOM), Maxlike (MAXL), Mahalanobis Distance (MAHA) and Radial Basis Function (RBF) had lower performance. We found inter-algorithm variations in predictor variable responses. We conclude that the multi-algorithm modelling approach enabled us to assess the variability in algorithm performance, hence a data driven basis for building the ensemble model. Given the inter-algorithm variations observed, the ensemble prediction maps indicated a better habitat utilization map of skipjack tuna than would have been achieved by a single algorithm.

Soil properties currently limiting crop yields in Swedish agriculture – An analysis of 90 yield survey districts and 10 long-term field experiments

• Among manageable soil properties, soil pH and extractable soil P had the largest impact on Swedish crop yields. • Current recommendations on soil pH and soil P require upward adjustment. • Surprisingly, increasing soil organic matter (SOM) reduced yields, most probably due to lower pH in soils richer in SOM. This study evaluated yield statistics and data from a comprehensive national soil inventory representing 90 yield survey districts and from the Swedish long-term soil fertility experiments. The aim was to identify the most yield-limiting soil properties and best agronomic practices in order to further improve yield. Principal Component Analysis, multi-variate analysis, step-wise regression, and tree partitioning analysis identified the same variables affecting yields in the national and experimental datasets. Crop yields were significantly affected by soil pH, soil organic matter (SOM), plant-available soil phosphorus (P-AL) and mean annual temperature. Contents of plant-available potassium and magnesium in soil had no significant impact on yield, except in potatoes. Soil pH was found to have the greatest potential to affect crop yields, even at values >6.5 (pH (H2O) ). Soil organic matter ranging from 3 to 6% in Swedish arable soils had an indirect negative effect on crop yields by lowering soil pH values with higher SOM content. To fully exploit the known benefits of SOM, liming requires more attention. One important finding was that current Swedish agricultural recommendations require updating. High-yielding crops demand more plant-available soil P and a range of 60−100 mg P-AL kg −1 soil is needed for sufficient P supply. A new target value of pH 7 for all crops except potatoes is recommended.

Untangling genotype x management interactions in multi-environment on-farm experimentation

• Identifying optimum crop design to match the environment remains a useful concept to maximise crop yields and farmers’ profits. • There is opportunity for field agronomic multi-environment experimentation to embrace system research approaches. • The value of applying a GxExM framework to increase yields and manage risks relates to our capacity to predict E at sowing and to the genetic diversity of the germplasm. Identifying optimum combinations of genotype (G) and agronomic management (M) i.e. crop design, to match the environment (E) i.e. site and expected seasonal conditions, is a useful concept to maximise crop yields and farmers’ profits. However, operationalising the concept requires practitioners to understand the likelihood of different E outcomes and GxM combinations that would maximise yields while managing risks. Here we propose and demonstrate an analysis framework to inform crop designs (GxM) at the time of sowing of a dryland maize crop, that combines data sets from multi-environment field experimentation and crop simulation modelling, and that accounts for risk preference. A network of replicated, G by M on-farm and on-research station trials (n = 10), conducted across New South Wales and Queensland, Australia, over three seasons (2014–2016) was collected. The trials consisted of combinations of commercial maize hybrids, sown at a range of plant densities and row configurations producing site average yields (Environment-yield) that varied between 1576 and 7914 kg ha −1 . Experimental data were used to test the capacity of APSIM-Maize 7.10 to simulate the experimental results, and to in-silico create a large synthetic data set of multi-E (sites x seasons) factorial combination of crop designs. Data mining techniques were applied on the synthetic data set, to derive a probabilistic model to predict the likely Environment-yield and associated risk from variables known at sowing, and to derive simple “rules of thumb” for farmers that discriminate high and low yielding crop designs across the lower, middle and upper tercile of the predicted Environment-yields. Four risk profiles are described, a “Dynamic” (i.e. each year the farmer would adopt a crop design based on the predicted Environment-yield tercile and corresponding “rules of thumb”), “High rewards seeker” (i.e. each year the farmer would adopt the crop design that optimises yield for the higher tercile of Environment-yields), “Middle’er” (i.e. each year the farmer would adopt the crop design that optimises yield for the middle tercile of Environment-yields), and “Risk averse” (i.e. each year the farmer would adopt the crop design that optimises yield for the lower tercile of Environment-yields). The difference in yield between the lowest and highest performing crop design was ca. 50 % which translates into a ca. 2-fold change in water use efficiency, i.e. from 8 to 15 kg grain mm −1 rainfall. APSIM-Maize explained 88 % of the variability in the experimental data set. The validated model was used to extend the number of E sampled by adding additional sites within the same region and using historical climate records for the period 1950–2018. Crop available water at the time of sowing was a good predictor for the likelihood of the season falling within each of the three Environment-yield terciles. Recursive partitioning trees showed that plant density and hybrid were the main variables discriminating crop performance within the upper, middle and lower terciles of Environment-yields. The probability distribution functions for yield resulting from the alternative risk management strategies were tested in terms of changes in the mean yield, an index of yield stability, and down-side risk i.e. the likelihood of achieving a non-economic yield. We conclude that (i) for dryland maize cropping, the crop water availability at the time of sowing can be used to inform optimum crop designs, increase yields and yield stability and reduce down-side risks; and (ii) the proposed framework is useful to untangle complex GxExM interactions in field experimentation that provide a transferable platform to develop simple rules to identify optimum crop designs early in the season.

On Scalability of Association-rule-based Recommendation

The association-rule-based approach is one of the most common technologies for building recommender systems and it has been extensively adopted for commercial use. A variety of techniques, mainly including eligible rule selection and multiple rules combination, have been developed to create effective recommendation. Unfortunately, little attention has been paid to the scalability concern of rule-based recommendation methods. However, the computational complexity of rule-based methods shall increase drastically with the growth of both online customers and rules, which are usually several millions in typical e-commerce platforms. Moreover, the dynamic change of users’ actions requires rule-based methods make recommendations in nearly real-time, which further highlights the scalability issue of rule-based recommender systems. In this article, we present a distributed framework that can scale different association-rule-based recommendation methods in a unified way. Specifically, based on the summarization of existing rule-based approaches, a generic tree-type structure is defined to store separate kinds of patterns, and an efficient algorithm is designed for mining eligible patterns along with computing recommendation scores. To handle the ever-increasing number of online customers, a distributed framework is proposed, where two load-balanced strategies for partitioning tree are put forward to fit sparse and dense data, respectively. Extensive experiments on five real-life data sets demonstrate that the efficiency of association-rule-based recommender systems can be significantly improved by the proposed framework.

A tree-based dictionary learning framework

We propose a new outline for adaptive dictionary learning methods for sparse encoding based on a hierarchical clustering of the training data. Through recursive application of a clustering method, the data is organized into a binary partition tree representing a multiscale structure. The dictionary atoms are defined adaptively based on the data clusters in the partition tree. This approach can be interpreted as a generalization of a discrete Haar wavelet transform. Furthermore, any prior knowledge on the wanted structure of the dictionary elements can be simply incorporated. The computational complexity of our proposed algorithm depends on the employed clustering method and on the chosen similarity measure between data points. Thanks to the multiscale properties of the partition tree, our dictionary is structured: when using Orthogonal Matching Pursuit to reconstruct patches from a natural image, dictionary atoms corresponding to nodes being closer to the root node in the tree have a tendency to be used with greater coefficients.

Efficient Scalable Video Coding Method Using Discrete Bandelet Transform

Video coding is an imperative part of the modern day communication system. Furthermore, it has vital roles in the fields of video streaming, multimedia, video conferencing and much more. Scalable Video Coding (SVC) is an emerging research area, due to its extensive application in most of the multimedia devices as well as public demand. The proposed coding technique is capable of eliminating the Spatio-temporal regularity of a video sequence. In Discrete Bandelet Transform (DBT), the directions are modeled by a three-directional vector field, known as structural flow. Regularity is decided by this flow where the data entropy is low. The wavelet vector decomposition of geometrically ordered data results in a lesser extent of significant coefficients. The directions of geometrical regularity are interpreted with a two-dimensional vector, and the approximation of these directions is found with spline functions. This paper deals with a novel SVC technique by exploiting the DBT. The bandelet coefficients are further encoded by utilizing Set Partitioning in Hierarchical Trees (SPIHT) encoder, followed by global thresholding mechanism. The proposed method is verified with several benchmark datasets using the performance measures which gives enhanced performance. Thus, the experimental results bring out the superiority of the proposed technique over the state-of-arts.