A Differential Approach for Gaze Estimation Liu, Gang; Yu, Yu; Mora, Kenneth A. Funes ...
IEEE transactions on pattern analysis and machine intelligence,
2021-March-1, 2021-Mar, 2021-3-1, 20210301, Letnik:
43, Številka:
3
Journal Article
Recenzirano
Odprti dostop
Most non-invasive gaze estimation methods regress gaze directions directly from a single face or eye image. However, due to important variabilities in eye shapes and inner eye structures amongst ...individuals, universal models obtain limited accuracies and their output usually exhibit high variance as well as subject dependent biases. Thus, increasing accuracy is usually done through calibration, allowing gaze predictions for a subject to be mapped to her actual gaze. In this article, we introduce a novel approach, which works by directly training a differential convolutional neural network to predict gaze differences between two eye input images of the same subject. Then, given a set of subject specific calibration images, we can use the inferred differences to predict the gaze direction of a novel eye sample. The assumption is that by comparing eye images of the same user, annoyance factors (alignment, eyelid closing, illumination perturbations) which usually plague single image prediction methods can be much reduced, allowing better prediction altogether. Furthermore, the differential network itself can be adapted via finetuning to make predictions consistent with the available user reference pairs. Experiments on 3 public datasets validate our approach which constantly outperforms state-of-the-art methods even when using only one calibration sample or those relying on subject specific gaze adaptation.
•We propose an interpretable partial differential network for SPLT-3D, where both horizontal and vertical interactions are modeled effectively.•We develop the theory of formulating 3D spatiotemporal ...dynamics with PDEs and design an effective strategy to apply it.•Experimental results show that our method can not only achieve better prediction performance, but also learn the correct PDEs.
Neural network-based methods have been widely applied to spatiotemporal prediction tasks, such as video prediction and weather forecasting. However, most existing works are designed for prediction in 2D space, and 3D prediction has not been extensively studied. In this paper, we propose to leverage 3D partial differential equations (PDEs) for spatiotemporal prediction in 3D space, and further develop a novel 3D neural partial differential network. This is inspired by that 3D PDEs can model both horizontal and vertical information interactions by various partial derivatives. Moreover, they can also formulate physical knowledge by equations. To integrate 3D PDEs in neural networks, we first develop the theory of approximating 3D partial derivatives by 3D convolutions, and further present an effective strategy to utilize the theory in practice. Then based on the theory and strategy, we propose a novel 3D Neural Partial Differential Network for prediction, named NPDN-3D. Specifically, NPDN-3D consists of two pivotal modules: (1) a neural partial differential module for capturing low-order spatiotemporal dynamics. This module is the key for prediction, where the dynamics are formulated by commonly-used low-order 3D PDEs. (2) A residual module for capturing the remaining non-low-order dynamics. This module performs as an extensible plug-in to enhance the expressiveness of our model. Extensive experiments on two simulated datasets and two real datasets show that our method not only achieves better prediction but also learns the correct PDEs.
Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease lacking effective treatments. This is due, in part, to a complex and incompletely understood pathophysiology. To shed light, ...we conducted untargeted metabolomics on plasma from two independent cross-sectional ALS cohorts versus control participants to identify recurrent dysregulated metabolic pathways. Untargeted metabolomics was performed on plasma from two ALS cohorts (cohort 1, n = 125; cohort 2, n = 225) and healthy controls (cohort 1, n = 71; cohort 2, n = 104). Individual differential metabolites in ALS cases versus controls were assessed by Wilcoxon, adjusted logistic regression and partial least squares-discriminant analysis, while group lasso explored sub-pathway level differences. Adjustment parameters included age, sex and body mass index. Metabolomics pathway enrichment analysis was performed on metabolites selected using the above methods. Additionally, we conducted a sex sensitivity analysis due to sex imbalance in the cohort 2 control arm. Finally, a data-driven approach, differential network enrichment analysis (DNEA), was performed on a combined dataset to further identify important ALS metabolic pathways. Cohort 2 ALS participants were slightly older than the controls (64.0 versus 62.0 years, P = 0.009). Cohort 2 controls were over-represented in females (68%, P < 0.001). The most concordant cohort 1 and 2 pathways centred heavily on lipid sub-pathways, including complex and signalling lipid species and metabolic intermediates. There were differences in sub-pathways that were enriched in ALS females versus males, including in lipid sub-pathways. Finally, DNEA of the merged metabolite dataset of both ALS and control cohorts identified nine significant subnetworks; three centred on lipids and two encompassed a range of sub-pathways. In our analysis, we saw consistent and important shared metabolic sub-pathways in both ALS cohorts, particularly in lipids, further supporting their importance as ALS pathomechanisms and therapeutics targets.
Pathway enrichment extensively used in the analysis of Omics data for gaining biological insights into the functional roles of pre-defined subsets of genes, proteins and metabolites. A large number ...of methods have been proposed in the literature for this task. The vast majority of these methods use as input expression levels of the biomolecules under study together with their membership in pathways of interest. The latest generation of pathway enrichment methods also leverages information on the topology of the underlying pathways, which as evidence from their evaluation reveals, lead to improved sensitivity and specificity. Nevertheless, a systematic empirical comparison of such methods is still lacking, making selection of the most suitable method for a specific experimental setting challenging. This comparative study of nine network-based methods for pathway enrichment analysis aims to provide a systematic evaluation of their performance based on three real data sets with different number of features (genes/metabolites) and number of samples.
The findings highlight both methodological and empirical differences across the nine methods. In particular, certain methods assess pathway enrichment due to differences both across expression levels and in the strength of the interconnectedness of the members of the pathway, while others only leverage differential expression levels. In the more challenging setting involving a metabolomics data set, the results show that methods that utilize both pieces of information (with NetGSA being a prototypical one) exhibit superior statistical power in detecting pathway enrichment.
The analysis reveals that a number of methods perform equally well when testing large size pathways, which is the case with genomic data. On the other hand, NetGSA that takes into consideration both differential expression of the biomolecules in the pathway, as well as changes in the topology exhibits a superior performance when testing small size pathways, which is usually the case for metabolomics data.
Background
With the rapid development and wide application of high-throughput sequencing technology, biomedical research has entered the era of large-scale omics data. We aim to identify genes ...associated with breast cancer prognosis by integrating multi-omics data.
Method
Gene-gene interactions were taken into account, and we applied two differential network methods JDINAC and LGCDG to identify differential genes. The patients were divided into case and control groups according to their survival time. The TCGA and METABRIC database were used as the training and validation set respectively.
Result
In the TCGA dataset, C11orf1, OLA1, RPL31, SPDL1 and IL33 were identified to be associated with prognosis of breast cancer. In the METABRIC database, ZNF273, ZBTB37, TRIM52, TSGA10, ZNF727, TRAF2, TSPAN17, USP28 and ZNF519 were identified as hub genes. In addition, RPL31, TMEM163 and ZNF273 were screened out in both datasets. GO enrichment analysis shows that most of these hub genes were involved in zinc ion binding.
Conclusion
In this study, a total of 15 hub genes associated with long-term survival of breast cancer were identified, which can promote understanding of the molecular mechanism of breast cancer and provide new insight into clinical research and treatment.
In the era of precision medicine, the analysis of simple information like sex and age can increase the potential to better diagnose and treat conditions that occur more frequently in one of the two ...sexes, present sex-specific symptoms and outcomes, or are characteristic of a specific age group. We present here a study of the association networks constructed from an array of 22 plasma metabolites measured on a cohort of 844 healthy blood donors. Through differential network analysis we show that specific association networks can be associated with sex and age: Different connectivity patterns were observed, suggesting sex-related variability in several metabolic pathways (branched-chain amino acids, ketone bodies, and propanoate metabolism). Reduction in metabolite hub connectivity was also found to be associated with age in both sex groups. Network analysis was complemented with standard univariate and multivariate statistical analysis that revealed age- and sex-specific metabolic signatures. Our results demonstrate that the characterization of metabolite–metabolite association networks is a promising and powerful tool to investigate the human phenotype at a molecular level.
Gene expression data provide an abundant resource for inferring connections in gene regulatory networks. While methodologies developed for this task have shown success, a challenge remains in ...comparing the performance among methods. Gold-standard datasets are scarce and limited in use. And while tools for simulating expression data are available, they are not designed to resemble the data obtained from RNA-seq experiments. SeqNet is an R package that provides tools for generating a rich variety of gene network structures and simulating RNA-seq data from them. This produces
RNA-seq data for benchmarking and assessing gene network inference methods. The package is available on CRAN and on GitHub at https://github.com/tgrimes/SeqNet.
Model organisms and human studies have yielded increasing empirical evidence that interactions among genes contribute broadly to genetic variation of complex traits. In the presence of gene-gene ...interactions, the dimensionality of the feature space becomes extremely high relative to the sample size. This poses a significant methodological challenge in the identification of gene-gene interactions. In this paper, by using a Gaussian graphical model framework, we translate the problem of identifying gene-gene interactions associated with a binary trait D into an inference problem on the difference of two high-dimensional precision matrices that summarize the conditional dependence network structures of the genes. We propose a procedure for testing the differential network globally, which is particularly powerful against sparse alternatives. In addition, a multiple testing procedure with false discovery rate control is developed to infer the specific structure of the differential network. Theoretical justification is provided to ensure the validity of the proposed tests, and optimality results are derived under sparsity assumptions. Through a simulation study we demonstrate that the proposed tests maintain the desired error rates under the null hypothesis and have good power under the alternative hypothesis. The methods are applied to a breast cancer gene expression study.