Novel machine learning insights into the QM7b and QM9 quantum mechanics datasets.

This paper (i) explores the internal structure of two quantum mechanics datasets (QM7b, QM9), composed of several thousands of organic molecules and described in terms of electronic properties, and (ii) further explores an inverse design approach to molecular design consisting of using machine learning methods to approximate the atomic composition of molecules, using QM9 data. Understanding the structure and characteristics of this kind of data is important when predicting the atomic composition from physical-chemical properties in inverse molecular designs. Intrinsic dimension analysis, clustering, and outlier detection methods were used in the study. They revealed that for both datasets the intrinsic dimensionality is several times smaller than the descriptive dimensions. The QM7b data is composed of well-defined clusters related to atomic composition. The QM9 data consists of an outer region predominantly composed of outliers, and an inner, core region that concentrates clustered inliner objects. A significant relationship exists between the number of atoms in the molecule and its outlier/inliner nature. The spatial structure exhibits a relationship with molecular weight. Despite the structural differences between the two datasets, the predictability of variables of interest for inverse molecular design is high. This is exemplified by models estimating the number of atoms of the molecule from both the original properties and from lower dimensional embedding spaces. In the generative approach the input is given by a set of desired properties of the molecule and the output is an approximation of the atomic composition in terms of its constituent chemical elements. This could serve as the starting region for further search in the huge space determined by the set of possible chemical compounds. The quantum mechanic's dataset QM9 is used in the study, composed of 133,885 small organic molecules and 19 electronic properties. Different multi-target regression approaches were considered for predicting the atomic composition from the properties, including feature engineering techniques in an auto-machine learning framework. High-quality models were found that predict the atomic composition of the molecules from their electronic properties, as well as from a subset of only 52.6% size. Feature selection worked better than feature generation. The results validate the generative approach to inverse molecular design.

Evaluation and Selection of Video Stabilization Techniques for UAV-Based Active Infrared Thermography Application.

Unmanned Aerial Vehicles (UAVs) that can fly around an aircraft carrying several sensors, e.g., thermal and optical cameras, to inspect the parts of interest without removing them can have significant impact in reducing inspection time and cost. One of the main challenges in the UAV based active InfraRed Thermography (IRT) inspection is the UAV's unexpected motions. Since active thermography is mainly concerned with the analysis of thermal sequences, unexpected motions can disturb the thermal profiling and cause data misinterpretation especially for providing an automated process pipeline of such inspections. Additionally, in the scenarios where post-analysis is intended to be applied by an inspector, the UAV's unexpected motions can increase the risk of human error, data misinterpretation, and incorrect characterization of possible defects. Therefore, post-processing is required to minimize/eliminate such undesired motions using digital video stabilization techniques. There are number of video stabilization algorithms that are readily available; however, selecting the best suited one is also challenging. Therefore, this paper evaluates video stabilization algorithms to minimize/mitigate undesired UAV motion and proposes a simple method to find the best suited stabilization algorithm as a fundamental first step towards a fully operational UAV-IRT inspection system.

Neural networks with multiple general neuron models: a hybrid computational intelligence approach using Genetic Programming.

Classical neural networks are composed of neurons whose nature is determined by a certain function (the neuron model), usually pre-specified. In this paper, a type of neural network (NN-GP) is presented in which: (i) each neuron may have its own neuron model in the form of a general function, (ii) any layout (i.e network interconnection) is possible, and (iii) no bias nodes or weights are associated to the connections, neurons or layers. The general functions associated to a neuron are learned by searching a function space. They are not provided a priori, but are rather built as part of an Evolutionary Computation process based on Genetic Programming. The resulting network solutions are evaluated based on a fitness measure, which may, for example, be based on classification or regression errors. Two real-world examples are presented to illustrate the promising behaviour on classification problems via construction of a low-dimensional representation of a high-dimensional parameter space associated to the set of all network solutions.

Multi-objective evolutionary optimization for constructing neural networks for virtual reality visual data mining: application to geophysical prospecting.

A method for the construction of virtual reality spaces for visual data mining using multi-objective optimization with genetic algorithms on nonlinear discriminant (NDA) neural networks is presented. Two neural network layers (the output and the last hidden) are used for the construction of simultaneous solutions for: (i) a supervised classification of data patterns and (ii) an unsupervised similarity structure preservation between the original data matrix and its image in the new space. A set of spaces are constructed from selected solutions along the Pareto front. This strategy represents a conceptual improvement over spaces computed by single-objective optimization. In addition, genetic programming (in particular gene expression programming) is used for finding analytic representations of the complex mappings generating the spaces (a composition of NDA and orthogonal principal components). The presented approach is domain independent and is illustrated via application to the geophysical prospecting of caves.

Time dependent neural network models for detecting changes of state in complex processes: applications in earth sciences and astronomy.

A computational intelligence approach is used to explore the problem of detecting internal state changes in time dependent processes; described by heterogeneous, multivariate time series with imprecise data and missing values. Such processes are approximated by collections of time dependent non-linear autoregressive models represented by a special kind of neuro-fuzzy neural network. Grid and high throughput computing model mining procedures based on neuro-fuzzy networks and genetic algorithms, generate: (i) collections of models composed of sets of time lag terms from the time series, and (ii) prediction functions represented by neuro-fuzzy networks. The composition of the models and their prediction capabilities, allows the identification of changes in the internal structure of the process. These changes are associated with the alternation of steady and transient states, zones with abnormal behavior, instability, and other situations. This approach is general, and its sensitivity for detecting subtle changes of state is revealed by simulation experiments. Its potential in the study of complex processes in earth sciences and astrophysics is illustrated with applications using paleoclimate and solar data.

Data mining of gene expression changes in Alzheimer brain.

Genome-wide transcription profiling is a powerful technique for studying the enormous complexity of cellular states. Moreover, when applied to disease tissue it may reveal quantitative and qualitative alterations in gene expression that give information on the context or underlying basis for the disease and may provide a new diagnostic approach. However, the data obtained from high-density microarrays is highly complex and poses considerable challenges in data mining. The data requires care in both pre-processing and the application of data mining techniques. This paper addresses the problem of dealing with microarray data that come from two known classes (Alzheimer and normal). We have applied three separate techniques to discover genes associated with Alzheimer disease (AD). The 67 genes identified in this study included a total of 17 genes that are already known to be associated with Alzheimer's or other neurological diseases. This is higher than any of the previously published Alzheimer's studies. Twenty known genes, not previously associated with the disease, have been identified as well as 30 uncharacterized expressed sequence tags (ESTs). Given the success in identifying genes already associated with AD, we can have some confidence in the involvement of the latter genes and ESTs. From these studies we can attempt to define therapeutic strategies that would prevent the loss of specific components of neuronal function in susceptible patients or be in a position to stimulate the replacement of lost cellular function in damaged neurons. Although our study is based on a relatively small number of patients (four AD and five normal), we think our approach sets the stage for a major step in using gene expression data for disease modeling (i.e. classification and diagnosis). It can also contribute to the future of gene function identification, pathology, toxicogenomics, and pharmacogenomics.