Elucidating Macular Structure-Function Correlations in Glaucoma

Correlation between structural data from optical coherence tomography (OCT) and functional data from the visual field (VF) may be suboptimal because of poor mapping of OCT measurement locations to VF test stimuli. We tested the hypothesis that stronger structure-function correlations in the macula can be achieved with fundus-tracking perimetery, by precisely mapping OCT measurements to VF sensitivity at the same location. The conventional 64 superpixel (3°x3°) OCT grid was mapped to VF sensitivities averaged in 40 corresponding VF units with standard automated perimetry (conventional mapped approach, CMA) in 38 glaucoma patients and 10 healthy subjects. Similarly, a 144 superpixel (2°x2°) OCT grid was mapped to each of the 68 VF locations with fundus-tracking perimetry (localized mapped approach, LMA). For each approach, the correlation between sensitivity at each VF unit and OCT superpixel was computed and the maximum value used to generate vector maps. CMA yielded significantly higher structure-function correlations compared to LMA. Only 20% of the vectors with CMA and <5% with LMA were within corresponding mapped OCT superpixels, while most were directed towards loci with structural damage. Measurement variability and patterns of glaucomatous damage are more likely to affect the correlations rather than precise mapping of VF stimuli.

Finding Similar Entities Across Knowledge Graphs

Finding similar entities among knowledge graphs is an essential research problem for knowledge integration and knowledge graph connection. This paper aims at finding semantically similar entities between two knowledge graphs. This can help end users and search agents more effectively and easily access pertinent information across knowledge graphs. Given a query entity in one knowledge graph, the proposed approach tries to find the most similar entity in another knowledge graph. The main idea is to leverage graph embedding, clustering, regression and sentence embedding. In this approach, RDF2Vec has been employed to generate vector representations of all entities of the second knowledge graph and then the vectors have been clustered based on cosine similarity using K medoids algorithm. Then, an artificial neural network with multilayer perception topology has been used as a regression model to predict the corresponding vector in the second knowledge graph for a given vector from the first knowledge graph. After determining the cluster of the predicated vector, the entities of the detected cluster are ranked through sentence-BERT method and finally the entity with the highest rank is chosen as the most similar one. To evaluate the proposed approach, experiments have been conducted on real-world knowledge graphs. The experimental results demonstrate the effectiveness of the proposed approach.

Mueller Matrix Polarimetric Imaging Analysis of Optical Components for the Generation of Cylindrical Vector Beams

In this work, we performed a Mueller matrix imaging analysis of two commercial optical components usually employed to generate and manipulate vector beams—a radial polarizer and a liquid-crystal q-plate. These two elements generate vector beams by different polarization mechanisms—polarizance and retardance, respectively. The quality of the vector beams relies on the quality of the device that generates them. Therefore, it is of interest to apply the well-established polarimetric imaging techniques to evaluate these optical components by identifying their spatial homogeneity in diattenuation, polarizance, depolarization, and retardance, as well as the spatial variation of the angles of polarizance and retardance vectors. For this purpose, we applied a customized imaging Mueller matrix polarimeter based on liquid-crystal retarders and a polarization camera. Experimental results were compared to the numerical simulations, considering the theoretical Mueller matrix. This kind of polarimetric characterization could be very helpful to the manufacturers and users of these devices.

Exploring Neural Network Hidden Layer Activity Using Vector Fields

Deep Neural Networks are known for impressive results in a wide range of applications, being responsible for many advances in technology over the past few years. However, debugging and understanding neural networks models’ inner workings is a complex task, as there are several parameters and variables involved in every decision. Multidimensional projection techniques have been successfully adopted to display neural network hidden layer outputs in an explainable manner, but comparing different outputs often means overlapping projections or observing them side-by-side, presenting hurdles for users in properly conveying data flow. In this paper, we introduce a novel approach for comparing projections obtained from multiple stages in a neural network model and visualizing differences in data perception. Changes among projections are transformed into trajectories that, in turn, generate vector fields used to represent the general flow of information. This representation can then be used to create layouts that highlight new information about abstract structures identified by neural networks.

Complex vectorial optics through gradient index lens cascades

Graded index (GRIN) lenses are commonly used for compact imaging systems. It is not widely appreciated that the ion-exchange process that creates the rotationally symmetric GRIN lens index profile also causes a symmetric birefringence variation. This property is usually considered a nuisance, such that manufacturing processes are optimized to keep it to a minimum. Here, rather than avoiding this birefringence, we understand and harness it by using GRIN lenses in cascade with other optical components to enable extra functionality in commonplace GRIN lens systems. We show how birefringence in the GRIN cascades can generate vector vortex beams and foci, and how it can be used advantageously to improve axial resolution. Through using the birefringence for analysis, we show that the GRIN cascades form the basis of a new single-shot Müller matrix polarimeter with potential for endoscopic label-free cancer diagnostics. The versatility of these cascades opens up new technological directions.

Training Temporal Word Embeddings with a Compass

Temporal word embeddings have been proposed to support the analysis of word meaning shifts during time and to study the evolution of languages. Different approaches have been proposed to generate vector representations of words that embed their meaning during a specific time interval. However, the training process used in these approaches is complex, may be inefficient or it may require large text corpora. As a consequence, these approaches may be difficult to apply in resource-scarce domains or by scientists with limited in-depth knowledge of embedding models. In this paper, we propose a new heuristic to train temporal word embeddings based on the Word2vec model. The heuristic consists in using atemporal vectors as a reference, i.e., as a compass, when training the representations specific to a given time interval. The use of the compass simplifies the training process and makes it more efficient. Experiments conducted using state-of-the-art datasets and methodologies suggest that our approach outperforms or equals comparable approaches while being more robust in terms of the required corpus size.

POSTER| Address2vec: Generating Vector Embeddings for Blockchain Analytics

Bitcoin Blockchain is a completely public open currency transaction ledger, recent growth in the crypto-currency market has driven many lawful and blackmarket actors using Bitcoin as the main method of payment. I propose address2vec an algorithm to generate vector representations of addresses on the Bitcoin Blockchain.I am able to present better results than a baseline random approach in predicting hoarding vs spending beaviour. The current work allows for utilization of common machine learning algorithms on bitcoin transaction addresses.

Drug repurposing through joint learning on knowledge graphs and literature

MotivationDrug repurposing is the problem of finding new uses for known drugs, and may either involve finding a new protein target or a new indication for a known mechanism. Several computational methods for drug repurposing exist, and many of these methods rely on combinations of different sources of information, extract hand-crafted features and use a computational model to predict targets or indications for a drug. One of the distinguishing features between different drug repurposing systems is the selection of features. Recently, a set of novel machine learning methods have become available that can efficiently learn features from datasets, and these methods can be applied, among others, to text and structured data in knowledge graphs.ResultsWe developed a novel method that combines information in literature and structured databases, and applies feature learning to generate vector space embeddings. We apply our method to the identification of drug targets and indications for known drugs based on heterogeneous information about drugs, target proteins, and diseases. We demonstrate that our method is able to combine complementary information from both structured databases and from literature, and we show that our method can compete with well-established methods for drug repurposing. Our approach is generic and can be applied to other areas in which multi-modal information is used to build predictive models.Availabilityhttps://github.com/bio-ontology-research-group/[email protected]

Semantic Disease Gene Embeddings (SmuDGE): phenotype-based disease gene prioritization without phenotypes

MotivationIn the past years, several methods have been developed to incorporate information about phenotypes into computational disease gene prioritization methods. These methods commonly compute the similarity between a disease’s (or patient’s) phenotypes and a database of gene-to-phenotype associations to find the phenotypically most similar match. A key limitation of these methods is their reliance on knowledge about phenotypes associated with particular genes which is highly incomplete in humans as well as in many model organisms such as the mouse.ResultsWe developed SmuDGE, a method that uses feature learning to generate vector-based representations of phenotypes associated with an entity. SmuDGE can be used as a trainable semantic similarity measure to compare two sets of phenotypes (such as between a disease and gene, or a disease and patient). More importantly, SmuDGE can generate phenotype representations for entities that are only indirectly associated with phenotypes through an interaction network; for this purpose, SmuDGE exploits background knowledge in interaction networks comprising of multiple types of interactions. We demonstrate that SmuDGE can match or outperform semantic similarity in phenotype-based disease gene prioritization, and furthermore significantly extends the coverage of phenotype-based methods to all genes in a connected interaction network.Availabilityhttps://github.com/bio-ontology-research-group/[email protected]

Emergent Yang–Mills theories from universal extra dimensions

We study emergent Yang–Mills theories which could origin from universal extra dimensions. Particularly, some vector field potential terms or polynomial vector field constraints introduced into five-dimensional (5D) non-Abelian gauge theory is shown to lead to spontaneous violation of an underlying spacetime symmetry and generate vector pseudo-Goldstone modes as conventional four-dimensional (4D) gauge boson candidates. As a special signature, apart from conventional gauge couplings, there appear an infinite number of the properly suppressed direct multi-boson (multi-photon in particular) interaction couplings in emergent Yang–Mills theories whose observation could shed light on their high-dimensional nature. Moreover, in these theories, an internal symmetry also appeared spontaneously broken to its diagonal subgroups. This breaking originates from the extra vector field components playing the role of some adjoint scalar field multiplet in the 4D spacetime. So, one naturally has the Higgs effect without a specially introduced scalar field multiplet. Remarkably, when applied to Grand Unified Theories (GUTs), this results in an automatic breakdown of emergent GUTs down to the Standard Model (SM) just at the 5D Lorentz violation scale M.