Dimensional Descriptors Research Articles

High-dimensional image descriptor matching using highly parallel KD-tree construction and approximate nearest neighbor search

To overcome the high computational cost associated with the high-dimensional digital image descriptor matching, this paper presents a set of integrated parallel algorithms for the construction of K-dimensional tree (KD-tree) and P approximate nearest neighbor search (P-ANNS) on the modern massively parallel architectures (MPA). To improve the runtime performance of the P-ANNS, we propose an efficient sliding window for a parallel buffered P-ANNS on KD-tree to mitigate the high cost of global memory accesses. When applied to high dimensional real-world image descriptor datasets, the proposed KD-tree construction and the buffered P-ANNS algorithms are of comparable matching quality to the traditional sequential counterparts on CPU, while outperforming their serial CPU counterparts by speedup factors of up to 17and 163, respectively. The algorithms are also studied for the performance impact factors to obtain the optimal runtime configurations for various datasets. Moreover, we verify the features of the parallel algorithms on typical 3D image matching scenarios. With the classical local image descriptor signature of histograms of orientations (SHOT) datasets, the parallel KD-tree construction and image descriptor matching can achieve up to 11 and 138-fold speedups, respectively.

A low dimensional descriptor for detection of anomalies in crowd videos

In this paper a novel descriptor is proposed for anomaly detection in crowd videos at a global level. Traditional approaches for anomaly detection in crowd videos face the dilemma of trade-off between high accuracy and real time performance. In order to resolve this issue, we propose an efficient descriptor composed of three different features extracted from the optical flow (OF) of a video sequence. The first feature is the sum of the optical flow field magnitude computed after applying a threshold. The second feature is the joint entropy of the OF magnitude of two consecutive frames used to measure the dissimilarity. The third feature is the variance computed from a space–time cuboid constructed using history of the OF field magnitude. Performance of the proposed descriptor is evaluated on the widely used UMN dataset in terms of accuracy and processing time. For UMN dataset, the proposed descriptor provides the highest area under the curve (AUC) compared to the approaches already published in literature.

Towards an alternative to nano-QSAR for nanoparticle toxicity ranking in case of small datasets

Statistical analysis approaches have been developed to predict the biological response to nanoparticles, especially quantitative structure–activity relationship (QSAR) models. But one major limitation remains the quantitative lack of data to build accurate models. The aim of this study was to investigate if simple alternative mathematical models could rank nanoparticles in a very binary way (i.e., toxic or not) in case of small dataset. We synthesized and characterized 25 nanoparticles from 6 metal (hydr)oxide families with particle size and shape tuning. We assessed their toxicity on RAW 264.7 cells and investigated relationships with both physicochemical and dimensional descriptors. A simple partial least square (PLS) regression analysis allowed ranking nanoparticles with respect to their toxicity, without false-negative results. But this model was not predictive due to the specific response of each family to dimensional parameters variations. A classification tree extracted the same main bulk descriptor as PLS, but interestingly showed the relevance of dimensional descriptors for the second and third node. We thus recommend the development of family-specific models and propose the combination of these two simple methods as pre-screening tools, a compromise to bridge the gap between case-by-case studies (expensive and time-consuming) and sophisticated nano-QSAR models (not suitable for small datasets).

Deep Learning to Speed up the Development of Structure-Property Relations For Hexagonal Boron Nitride and Graphene.

Structure-property maps play a key role in accelerated materials discovery. The current norm for developing these maps includes computationally expensive physics-based simulations. Here, the capabilities of deep learning agents are explored such as convolutional neural networks (CNNs) and multilayer perceptrons (MLPs) to predict structure-property relations and reduce dependence on simulations. This study contains simulated hexagonal boron nitride (h-BN) microstructures damaged by various levels of radiation and temperature, with the objective to predict their residual strengths from the final atomic positions. By developing low dimensional physical descriptors to statistically describe the defects, these results show that purpose-specific microstructure representation can help in achieving a good prediction accuracy at low computational cost. Furthermore, the adaptability of the trained deep learning agents is explored to predict structure-property maps of other 2D materials using transfer learning. It is shown that in order to achieve good predictions accuracy (≈95% R2 ), an agent that is training for the first time ("learning from scratch") requires 23-45% of simulated data, whereas an agent adapting to a different material ("transfer learning") requires only about 10% or less. This suggests that transfer learning is a potential game changer in material discovery and characterization approaches.

Music Visualization Based on Spherical Projection With Adjustable Metrics

Development of techniques for music visualization is important and still open problem in analysis and creation of the quantitative profiles of single or multiple compositions, which could be used as required constraints in music generation or music classification processes. When generating creative data with no objective function, it is hard to select or to find appropriate measurable features. This paper proposes a method to normalize data in MIDI files by 12 dimensional vector descriptors extracted from tonality as well as a novel technique for dimensionality reduction and visualization of extracted music data by 3D projections is discussed. Employing a non-overlapping sliding window through the composition, the harmonic features are found in a music piece. Then a self-similarity matrix is computed using distance metrics to analyze and project the resulting 3D feature vectors. Three dimensional projection creates a quantitative profile of a composition, which correlates the tone similarities along the music piece. The dimensionality reduction is compared with well-known autoencoder. Conducted tests show that our method preserves up to 90% of original data in the projection of reduced dimension. The advantages of the proposed method consist in a novel technique that provides interactive visualization and dynamically adjusts different metrics to observe the behavior of data during music information retrieval and recognition.

Visual Object Tracking Using Structured Sparse PCA-Based Appearance Representation and Online Learning.

Visual object tracking is a fundamental research area in the field of computer vision and pattern recognition because it can be utilized by various intelligent systems. However, visual object tracking faces various challenging issues because tracking is influenced by illumination change, pose change, partial occlusion and background clutter. Sparse representation-based appearance modeling and dictionary learning that optimize tracking history have been proposed as one possible solution to overcome the problems of visual object tracking. However, there are limitations in representing high dimensional descriptors using the standard sparse representation approach. Therefore, this study proposes a structured sparse principal component analysis to represent the complex appearance descriptors of the target object effectively with a linear combination of a small number of elementary atoms chosen from an over-complete dictionary. Using an online dictionary for learning and updating by selecting similar dictionaries that have high probability makes it possible to track the target object in a variety of environments. Qualitative and quantitative experimental results, including comparison to the current state of the art visual object tracking algorithms, validate that the proposed tracking algorithm performs favorably with changes in the target object and environment for benchmark video sequences.

Modeling Detection Statistics in Feature‐Based Robotic Navigation for Range Sensors

This paper proposes using the number of range measurements that a detector utilizes to generate a detection as its descriptor. This one dimensional descriptor can be calculated with many range-based detectors, and its expected value is used to derive detection statistics which take into account feature occlusions to improve robotic navigation performance. To demonstrate the advantages of estimating detection statistics, they are estimated and tested within Random Finite Set and vector-based Simultaneous Localization and Mapping (SLAM) algorithms. Results from simulations and real experiments demonstrate the advantages of explicitly modeling feature detection statistics in both frameworks. © 2018 Institute of Navigation.

A review of the concepts of ‘axony’ and their bearing on tetrapod ichnology

ABSTRACT Two meanings of the term axony are found in the ichnological literature. Multiple meanings may prove to be a double-edge sword, complicating scientific communication. In vertebrate ichnology the first meaning of axony relies on aspects of locomotion related to the body weight support and propulsive thrust. A second one concerns axony as a purely geometric and dimensional descriptor. These approaches are based on a static view of the impression process, implying the loss of much important information. Here we report an analysis of shallowly impressed footprints referred to the ichnotaxa Ichniotherium sphaerodactylum and Dimetropus osageorum. The analysis was carried out by considering the track registration as a dynamic process and attempting to identify and describe axony conditions during movements. Variations in the axony conditions can be understood in the light of the producer’s foot anatomy and the reciprocal relations between foot bone elements. The concept of axony can be a useful tool in ichnological practice only when it is related to the complex dynamic of locomotion and the resulting track registration. It can help in restoring the interconnections between track and trackmaker, re-establishing the biological significance of tetrapod footprints.

Synthesis and SAR of new isoxazole-triazole bis-heterocyclic compounds as analogues of natural lignans with antiparasitic activity

Despite the impressive scientific and technological advances of recent decades, no effective treatment is currently available for Chagas disease. Our research group has been studying the design and synthesis of analogues of natural lignans aiming to identify compounds with antiparasitic activity. This article reports the synthesis of 42 novel bis-heterocyclic derivatives and the structure-activity relationship study conducted based on results of biological assays against Trypanosoma cruzi amastigotes. Thirty-seven compounds were active, and eight of them had GI50 values lower than 100 μM (GI50 88.4–12.2 μM). A qualitative structure activity relationship study using three dimensional descriptors was carried out and showed a correlation between growth inhibitory potency and the presence of bulky hydrophobic groups located at rings A and D of the compounds. Compound 3-(3,4-dimethoxyphenyl)-5-((4-(4-pentylphenyl)-1H-1,2,3-triazol-1-yl)methyl)isoxazole (31) was the most active in the series (GI50 12.2 μM), showing, in vitro, low toxicity and potency similar to benznidazole (GI50 10.2 μM). These results suggest that this compound can be a promising scaffold for the design of new trypanocidal compounds.

Two-Dimensional Canonical Correlation Analysis of the Logically Concatenated Cross Binary Pattern for Cross Pose Face Recognition

Face pose recognition is one of the challenging areas in computer vision. Cross-pose change causes the change in the information of face appearance. The maximization of intrasubject correlation helps to widen the intersubject differences which helps further in achieving pose invariance. In this paper, for cross pose recognition, the authors propose to maximize the cross pose correlation by using the logically concatenated cross binary pattern (LC-CBP) descriptor and two dimensional canonical correlation analysis (2DCCA). The LC-CBP descriptor extracts the local texture details of face images with low computation complexity and the 2DCCA explicitly searches for the maximization of the correlated features to retain most informative content. Joint feature consideration via 2DCCA helps in setting up a better correspondence between a discrete set of nonfrontal pose and the frontal pose of the same subject. Experimental results demonstrate the two dimensional canonical correlation LC-CBP descriptor along with intensity values improve the correlation.

Optimized peer to peer QSPR prediction of enthalpy of formation using outlier detection and subset selection

Quantitative Structure Property Relationship (QSPR) approach provides a model to understand the property or activity of a molecule by identifying the relationship with its chemical structure. Accurate identification of property of a molecule has higher influence of modern drug discovery system. Hence, development of an efficient method to identify the molecule property or activity is becoming mandatory component in drug design. Modern drug discovery system involves cluster networks since drug design process can be enhanced with the support of peer to peer networks. Few Molecule will have high dimensional structure and descriptor information where these information can be efficiently handled by cluster networks. This research work visualizes existing bench marking models based on Polynomial Regression (PR), Principal Component Regression (PCR) and Partial Least Square Regression (PLSR) with respect to fitted response and prediction. An optimized QSPR model (FDROL) with fuzzy minimum redundancy maximum relevance (FmRMR) data reduction (FDR) and outlier detection (OL) was proposed. The influences of topological descriptor to predict the physicochemical property of hydrocarbons have been determined. The dataset has been analyzed by proposed method using polynomial regression (PRFDROL), principal component regression (PCFDROL) and partial least square regression (PLFDROL) to predict the enthalpy of formation of hydrocarbons. The model was validated with high correlation coefficient (r, r2, adjr2, F) and lower standard error (se) which shows that the model has good predictive ability. The squared correlation coefficient (r2) for preprocessed data using PR, PCR and PLSR were obtained as 1, 0.98392 and 0.9839 which were better predicted and fitted responses compare with existing methods. The optimized QSPR model with PR shows best fit to predict enthalpy of formation of hydrocarbons.

INTEGRATING STRUCTURE AND LIGAND-BASED APPROACHES FOR MODELLING THE HISTONE DEACETYLASE INHIBITION ACTIVITY OF HYDROXAMIC ACID DERIVATIVES

Objective: Structure and ligand-based drug design approaches have be been integrated to accurately predict the inhibition activity of hydroxamic acid (HA) derivatives against the histone deacetylase-2 enzyme (HDAC2).Methods: The “active conformations” of the ligands in the binding site of the enzyme were determined by docking assays. More than 1000 0–3 dimensional molecular descriptors included in Dragon package were calculated and utilized for developing quantitative structure-activity relationship (QSAR) models through a multiple linear regression approach coupled with the genetic algorithm (GA-MLR).Results: The final model obtained showed suitable robustness and stability, with low correlation between descriptors and good predictive power. QSAR model was then used for screening bioactivity from a series of 36 novel HAs and found five candidates with very good bioactivity (half maximal inhibitory concentration<0.1 μM). Docking experiment revealed the binding mode of these compounds into the active site of HDAC2. Drug-likeness and toxicity profiles of the compounds were checked through chemoinformatics tools.Conclusion: The results from this study can lead to rational design and synthesis of highly selective and potent HDAC2 inhibitors.

MOLTRA-II. New three dimensional descriptors of the hydrogen bond

The program MOLTRA-II, created on the basis of the CHEmical Databases management system CHED, is supplemented by the possibility of calculating new descriptors based on the three-dimensional structure and hydrogen bond parameters. The program calculates PSA descriptors (Polar Surface Area), OSA (Optimal Structure Area) and energy of the hydrogen bond.

Probabilistic modeling of anatomical variability using a low dimensional parameterization of diffeomorphisms.

We present an efficient probabilistic model of anatomical variability in a linear space of initial velocities of diffeomorphic transformations and demonstrate its benefits in clinical studies of brain anatomy. To overcome the computational challenges of the high dimensional deformation-based descriptors, we develop a latent variable model for principal geodesic analysis (PGA) based on a low dimensional shape descriptor that effectively captures the intrinsic variability in a population. We define a novel shape prior that explicitly represents principal modes as a multivariate complex Gaussian distribution on the initial velocities in a bandlimited space. We demonstrate the performance of our model on a set of 3D brain MRI scans from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. Our model yields a more compact representation of group variation at substantially lower computational cost than the state-of-the-art method such as tangent space PCA (TPCA) and probabilistic principal geodesic analysis (PPGA) that operate in the high dimensional image space.

On Mixed Galerkin Modeling for Controller Design of Boundary Control Systems

A mixed Galerkin formulation for reduced-order modeling is presented for controller design of distributed parameter systems with boundary inputs. A null space approach is established to yield appropriate trial and test functions by using candidate trial functions which are free from boundary conditions. The approach is applicable, with considerably mild conditions on the candidate trial functions, on the interior domain as well as its boundaries. It is shown that this modeling technique yields a finite dimensional descriptor system of index 1, that is equivalent to a state-space representation suitable for controller design model. Effectiveness of the proposed method is demonstrated by an example of a heat conduction rod using polynomial functions which are not preprocessed to meet the boundary conditions.

Three-Dimensional Biologically Relevant Spectrum (BRS-3D): Shape Similarity Profile Based on PDB Ligands as Molecular Descriptors.

The crystallized ligands in the Protein Data Bank (PDB) can be treated as the inverse shapes of the active sites of corresponding proteins. Therefore, the shape similarity between a molecule and PDB ligands indicated the possibility of the molecule to bind with the targets. In this paper, we proposed a shape similarity profile that can be used as a molecular descriptor for ligand-based virtual screening. First, through three-dimensional (3D) structural clustering, 300 diverse ligands were extracted from the druggable protein–ligand database, sc-PDB. Then, each of the molecules under scrutiny was flexibly superimposed onto the 300 ligands. Superimpositions were scored by shape overlap and property similarity, producing a 300 dimensional similarity array termed the “Three-Dimensional Biologically Relevant Spectrum (BRS-3D)”. Finally, quantitative or discriminant models were developed with the 300 dimensional descriptor using machine learning methods (support vector machine). The effectiveness of this approach was evaluated using 42 benchmark data sets from the G protein-coupled receptor (GPCR) ligand library and the GPCR decoy database (GLL/GDD). We compared the performance of BRS-3D with other 2D and 3D state-of-the-art molecular descriptors. The results showed that models built with BRS-3D performed best for most GLL/GDD data sets. We also applied BRS-3D in histone deacetylase 1 inhibitors screening and GPCR subtype selectivity prediction. The advantages and disadvantages of this approach are discussed.

A Bayesian network model for predicting aquatic toxicity mode of action using two dimensional theoretical molecular descriptors

The mode of toxic action (MoA) has been recognized as a key determinant of chemical toxicity, but development of predictive MoA classification models in aquatic toxicology has been limited. We developed a Bayesian network model to classify aquatic toxicity MoA using a recently published dataset containing over one thousand chemicals with MoA assignments for aquatic animal toxicity. Two dimensional theoretical chemical descriptors were generated for each chemical using the Toxicity Estimation Software Tool. The model was developed through augmented Markov blanket discovery from the dataset of 1098 chemicals with the MoA broad classifications as a target node. From cross validation, the overall precision for the model was 80.2%. The best precision was for the AChEI MoA (93.5%) where 257 chemicals out of 275 were correctly classified. Model precision was poorest for the reactivity MoA (48.5%) where 48 out of 99 reactive chemicals were correctly classified. Narcosis represented the largest class within the MoA dataset and had a precision and reliability of 80.0%, reflecting the global precision across all of the MoAs. False negatives for narcosis most often fell into electron transport inhibition, neurotoxicity or reactivity MoAs. False negatives for all other MoAs were most often narcosis. A probabilistic sensitivity analysis was undertaken for each MoA to examine the sensitivity to individual and multiple descriptor findings. The results show that the Markov blanket of a structurally complex dataset can simplify analysis and interpretation by identifying a subset of the key chemical descriptors associated with broad aquatic toxicity MoAs, and by providing a computational chemistry-based network classification model with reasonable prediction accuracy.

Preliminary studies on sensory, instrumental and chemical evaluation of dried goji (Lycium barbarumL.) berries

Six commercial brands of dried goji (Lycium barbarum L.) berries were evaluated for their sensory, physical and chemical parameters. A quantitative descriptive analysis (QDA with 18 attributes) was undertaken and consumer preferences determined. Sensory assessments were supported with physical measurements of appearance and texture. Chemical parameters included: total phenolic content, some phenolic acids (chlorogenic, caffeic, p-cumaric, ferulic and sinapic acids), antioxidant activity and ascorbic acid content. The berries were analyzed for their mineral elements (Ca, Cr, Fe, Ge, K, Na, P, Se, Zn) content. The main aim of this work was to assess differences between dried goji berries brands available in the Italian market and to relate them with the data available in the literature. Berries showed few differences in sensory, instrumental and chemical parameters, and those uncovered mainly related to visual, dimensional and texture descriptors (sensory analysis). Dried goji berries were not particularly appreciated by consumers and this suggests that they are bought for their advertised medicinal properties. For the most relevant compounds (e.g., phenolic acids, antioxidant molecules and elements), only a few variations were observed among brands. Dried goji berries showed to be an excellent source of antioxidants and mineral elements able to contribute in a relevant matter to the recommended daily intakes.

Out-of-Sample Extrapolation utilizing Semi-Supervised Manifold Learning (OSE-SSL): Content Based Image Retrieval for Histopathology Images.

Content-based image retrieval (CBIR) retrieves database images most similar to the query image by (1) extracting quantitative image descriptors and (2) calculating similarity between database and query image descriptors. Recently, manifold learning (ML) has been used to perform CBIR in a low dimensional representation of the high dimensional image descriptor space to avoid the curse of dimensionality. ML schemes are computationally expensive, requiring an eigenvalue decomposition (EVD) for every new query image to learn its low dimensional representation. We present out-of-sample extrapolation utilizing semi-supervised ML (OSE-SSL) to learn the low dimensional representation without recomputing the EVD for each query image. OSE-SSL incorporates semantic information, partial class label, into a ML scheme such that the low dimensional representation co-localizes semantically similar images. In the context of prostate histopathology, gland morphology is an integral component of the Gleason score which enables discrimination between prostate cancer aggressiveness. Images are represented by shape features extracted from the prostate gland. CBIR with OSE-SSL for prostate histology obtained from 58 patient studies, yielded an area under the precision recall curve (AUPRC) of 0.53 ± 0.03 comparatively a CBIR with Principal Component Analysis (PCA) to learn a low dimensional space yielded an AUPRC of 0.44 ± 0.01.

Chain coding representation of voxel-based objects with enclosing, edging and intersecting trees

Thrifty methods to represent and store three dimensional objects are important. Two different methods for describing voxel-based objects (VBOs) by means of edging (ETs) and intersecting (ITs) trees are demonstrated. Each tree comes from a different kind of border of the underlying VBO, and both trees are one dimensional alternative descriptors to skeletons for VBOs representation. Vertices in the trees correspond to the vertices of the VBO enclosing surface where some surface vertices have been conveniently suppressed. These descriptors are computed using a base-five digit chain code (combined with parentheses) and has been used to illustrate three dimensional curves and enclosing trees. The descriptors are invariant under rotation and translation, and preserve the VBO shape. Using either descriptor, the description of the mirror image of a VBO is easily obtained. The proposed descriptor notation is a good tool for storing VBOs, and intersecting trees providing further storage savings. Enclosing trees (EcTs) are briefly reviewed as a preamble to introduce ETs and ITs.