Multi-Scale Memetic Image Registration

Many technological applications of our time rely on images captured by multiple cameras. Such applications include the detection and recognition of objects in captured images, the tracking of objects and analysis of their motion, and the detection of changes in appearance. The alignment of images captured at different times and/or from different angles is a key processing step in these applications. One of the most challenging tasks is to develop fast algorithms to accurately align images perturbed by various types of transformations. The paper reports a new method used to register images in the case of geometric perturbations that include rotations, translations, and non-uniform scaling. The input images can be monochrome or colored, and they are preprocessed by a noise-insensitive edge detector to obtain binarized versions. Isotropic scaling transformations are used to compute multi-scale representations of the binarized inputs. The algorithm is of memetic type and exploits the fact that the computation carried out in reduced representations usually produces promising initial solutions very fast. The proposed method combines bio-inspired and evolutionary computation techniques with clustered search and implements a procedure specially tailored to address the premature convergence issue in various scaled representations. A long series of tests on perturbed images were performed, evidencing the efficiency of our memetic multi-scale approach. In addition, a comparative analysis has proved that the proposed algorithm outperforms some well-known registration procedures both in terms of accuracy and runtime.

A journey through mapping space: characterising the statistical and metric properties of reduced representations of macromolecules

A mapping of a macromolecule is a prescription to construct a simplified representation of the system in which only a subset of its constituent atoms is retained. As the specific choice of the mapping affects the analysis of all-atom simulations as well as the construction of coarse-grained models, the characterisation of the mapping space has recently attracted increasing attention. We here introduce a notion of scalar product and distance between reduced representations, which allows the study of the metric and topological properties of their space in a quantitative manner. Making use of a Wang–Landau enhanced sampling algorithm, we exhaustively explore such space, and examine the qualitative features of mappings in terms of their squared norm. A one-to-one correspondence with an interacting lattice gas on a finite volume leads to the emergence of discontinuous phase transitions in mapping space, which mark the boundaries between qualitatively different reduced representations of the same molecule. Graphicabstract

Gaussian process analysis of electron energy loss spectroscopy data: multivariate reconstruction and kernel control

Advances in hyperspectral imaging including electron energy loss spectroscopy bring forth the challenges of exploratory and physics-based analysis of multidimensional data sets. The multivariate linear unmixing methods generally explore similarities in the energy dimension, but ignore correlations in the spatial domain. At the same time, Gaussian process (GP) explicitly incorporate spatial correlations in the form of kernel functions but is computationally intensive. Here, we implement a GP method operating on the full spatial domain and reduced representations in the energy domain. In this multivariate GP, the information between the components is shared via a common spatial kernel structure, while allowing for variability in the relative noise magnitude or image morphology. We explore the role of kernel constraints on the quality of the reconstruction, and suggest an approach for estimating them from the experimental data. We further show that spatial information contained in higher-order components can be reconstructed and spatially localized.

Mash Sketched Reference Dataset for Genome-Based Taxonomy and Comparative Genomics

The analysis of curated genomic, metagenomic, and proteomic data are of paramount importance in the fields of biology, medicine, education, and bioinformatics. Although this type of data is usually hosted in raw form in free international repositories, its access requires plenty of computing, storage, and processing capacities for the domestic user. The purpose of the study is to offer a comprehensive set of genomic and proteomic reference data, in an accessible and easy-to-use form to the scientific community. A representative type material set of genomes, proteomes and metagenomes were directly downloaded from the site: https://www.ncbi.nlm.nih.gov/assembly/ and from Genome Taxonomy Database, associated with the major groups of Bacteria, Archaea, Virus, and Fungi. Sketched databases were subsequently created and stored on handy raw reduced representations, by using Mash software. Our dataset contains near to 100 GB of space disk reduced to 585.78 MB and represents 87,476 genomics/proteomic records from eight informative contexts, which have been prefiltered to make them accessible, usable, and user-friendly with computational resources. Potential uses of this dataset include but are not limited to, microbial species delimitation, estimation of genomic distances, genomic novelties, paired comparisons between proteomes, genomes, and metagenomes.

A Deep Graph Network–Enhanced Sampling Approach to Efficiently Explore the Space of Reduced Representations of Proteins

The limits of molecular dynamics (MD) simulations of macromolecules are steadily pushed forward by the relentless development of computer architectures and algorithms. The consequent explosion in the number and extent of MD trajectories induces the need for automated methods to rationalize the raw data and make quantitative sense of them. Recently, an algorithmic approach was introduced by some of us to identify the subset of a protein’s atoms, or mapping, that enables the most informative description of the system. This method relies on the computation, for a given reduced representation, of the associated mapping entropy, that is, a measure of the information loss due to such simplification; albeit relatively straightforward, this calculation can be time-consuming. Here, we describe the implementation of a deep learning approach aimed at accelerating the calculation of the mapping entropy. We rely on Deep Graph Networks, which provide extreme flexibility in handling structured input data and whose predictions prove to be accurate and-remarkably efficient. The trained network produces a speedup factor as large as 105 with respect to the algorithmic computation of the mapping entropy, enabling the reconstruction of its landscape by means of the Wang–Landau sampling scheme. Applications of this method reach much further than this, as the proposed pipeline is easily transferable to the computation of arbitrary properties of a molecular structure.

LinCoM: a graph theoretic approach for the production of Linear place fields in Complex Mazes

Studies on spatial coding and episodic memory typically involve recordings of hippocampal place cell activity while rodents navigate in mazes. Linear place fields serve as reduced representations of the activity of place cells, revealing their spatial preference along the tracks of the maze. Sometimes, the experimental designs include complex mazes with irregular geometries and one or more decision points. Unfortunately, in such complex mazes, the production of linear place fields becomes a non-trivial problem. Here, I present a MATLAB toolbox which implements a graph-theoretic approach for the efficient production of linear place fields in a variety of complex mazes.

Co-Simulation of Multibody Systems With Contact Using Reduced Interface Models

Co-simulation techniques enable the coupling of physically diverse subsystems in an efficient and modular way. Communication between subsystems takes place at discrete-time instants and is limited to a given set of coupling variables, while the internals of each subsystem remain undisclosed and are generally not accessible to the rest of the simulation environment. In noniterative co-simulation schemes, commonly used in real-time applications, this may lead to the instability of the numerical integration. The stability of the integration in these cases can be enhanced using interface models, i.e., reduced representations of one or more subsystems that provide physically meaningful input values to the other subsystems between communication points. This work describes such an interface model that can be used to represent nonsmooth mechanical systems subjected to unilateral contact and friction. The dynamics of the system is initially formulated as a mixed linear complementarity problem (MLCP), from which the effective mass and force terms of the interface model are derived. These terms account for contact detachment and stick–slip transitions, and can also include constraint regularization in case of redundancy in the system. The performance of the proposed model is shown in several challenging examples of noniterative multirate co-simulation schemes of a mechanical system with hydraulic components, which feature faster dynamics than the multibody subsystem. Using an interface model improves simulation stability and allows for larger integration step-sizes, thus resulting in a more efficient simulation.

Foraging fruit flies mix navigational and learning-based decision-making strategies

Animals often navigate environments that are uncertain, volatile and complex, making it challenging to locate reliable food sources. Therefore, it is not surprising that many species evolved multiple, parallel and complementary foraging strategies to survive. Current research on animal behavior is largely driven by a reductionist approach and attempts to study one particular aspect of behavior in isolation. This is justified by the huge success of past and current research in understanding neural circuit mechanisms of behaviors. But focusing on only one aspect of behaviors obscures their inherent multidimensional nature. To fill this gap we aimed to identify and characterize distinct behavioral modules using a simple reward foraging assay. For this we developed a single-animal, trial-based probabilistic foraging task, where freely walking fruit flies experience optogenetic sugar-receptor neuron stimulation. By carefully analyzing the walking trajectories of flies, we were able to dissect the animals foraging decisions into multiple underlying systems. We show that flies perform local searches, cue-based navigation and learn task relevant contingencies. Using probabilistic reward delivery allowed us to bid several competing reinforcement learning (RL) models against each other. We discover that flies accumulate chosen option values, forget unchosen option values and seek novelty. We further show that distinct behavioral modules -learning and navigation-based systems-cooperate, suggesting that reinforcement learning in flies operates on dimensionality reduced representations. We therefore argue that animals will apply combinations of multiple behavioral strategies to generate foraging decisions.

Encoding structure in holographic reduced representations

Vector Symbolic Architectures (VSAs) such as Holographic Reduced Representations (HRRs) are computational associative memories used by cognitive psychologists to model behavioural and neurological aspects of human memory. We present a novel analysis of the mathematics of VSAs and a novel technique for representing data in HRRs. Encoding and decoding in VSAs can be characterized by Latin squares. Successful encoding requires the structure of the data to be orthogonal to the structure of the Latin squares. However, HRRs can successfully encode vectors of locally structured data if vectors are shuffled. Shuffling results are illustrated using images, but are applicable to any non-random data. The ability to use locally structured vectors provides a technique for detailed modelling of stimuli in HRR models.

Interface Models for Multirate Co-Simulation of Nonsmooth Multibody Systems

Co-simulation techniques enable the coupling of physically diverse subsystems in an efficient and modular way. Complex engineering applications can be simulated in co-simulation setups, in which each subsystem is solved and integrated using numerical methods tailored to its physical behaviour. Co-simulation implies that the communication between subsystems takes place at discrete-time instants and is limited to a given set of coupling variables, while the internals of each subsystem are generally not accessible to the rest of the simulation environment. In non-iterative co-simulation schemes, this may lead to the instability of the integration. Increasingly demanding requirements in the simulation of machinery have led to the coupling, in real-time co-simulation setups, of multibody models of mechanical systems to computational representations of non-mechanical subsystems, such as hydraulics and electronics. Often, these feature faster dynamics than their mechanical counterparts, which leads to the use of multirate integration in non-iterative co-simulation environments. The stability of the integration in these cases can be enhanced using interface models, i.e., reduced representations of the multibody system, to provide meaningful input values to faster subsystems between communication points. This work describes such interface models that can be used to represent nonsmooth mechanical systems subjected to unilateral contact and friction.