Shlomit Gur


PATENet: Pairwise Alignment of Time Evolving Networks

Gur S., and Honava V.J.
in Machine Learning and Data Mining in Pattern Recognition [MLDM 2018]
Key words: Network science, Multilayer networks, Temporal alignment

Networks that change over time, e.g. functional brain networks that change their structure due to processes such as development or aging, are naturally modeled by time-evolving networks. In this paper we present PATENet, a novel method for aligning time-evolving networks. PATENet offers a mathematically-sound approach to aligning time evolving networks. PATENet leverages existing similarity measures for networks with fixed topologies to define well-behaved similarity measures for time evolving networks. We empirically explore the behavior of PATENet through synthetic time evolving networks under a variety of conditions.
Python implementation is avaiable at:
Age-Related Differences in Task-Specific Functional Connectivity in Phonological and Semantic
Picture-Based Match-Mismatch Tasks in the Presence of Task-Irrelevant Written Words

Gur S., El-Manzalawy Y., Diaz M.T., and Honava V.J.
[Under Review]
Key words: functional connectivity, aging, semantics, phonology, machine learning, attention

Language-related cognitive processes are affected differently by aging: phonological processes display
deficits, whereas semantic processes deficits are minimal. The inhibition deficit hypothesis of aging suggests that older adults find it more difficult to ignore irrelevant information compared to younger adults. We investigated age-group-related/independent differences between semantic and phonological processes, using functional connectivity from younger and older adults engaged in picture-based phonological and semantic judgment tasks in the presence of task-irrelevant written words. The words were phonologically or
semantically related to the pictures in the phonological and semantic trials, respectively. Specifically, we characterized the task-specific functional connectivity networks using network topological features, and used statistical machine learning algorithms to train predictive models to reliably make semantic versus phonological decisions from these data.

We used the resulting models to identify network topological features that reliably discriminate between functional connectivity measured during the two tasks in younger and older adults separately, as well as combined together. We find that the left thalamus and the left caudate nucleus show greater functional connectivity in the phonological task, as compared to the semantic task,
regardless of age. Our results show greater similarity in functional connectivity between the
semantic and phonological tasks in older adults as compared to younger adults, a result that is consistent with the dedifferentiation hypothesis.
We also find age related differences in the functional connectivity of specific brain regions between the semantic and phonological tasks (greater functional connectivity of right hippocampus in the semantic task in younger adults, and greater functional connectivity of
left superior parietal lobule and brainstem in the phonological task in older adults).

Our results suggest that the functional connectivity induced by semantic and phonological processes differs across brain regions that have been implicated in intrinsic alertness, a prerequisite for selective attention to aspects of the tasks. Our results also demonstrate the potential of data-driven machine learning algorithms to contribute evidence to support, refute, or even suggest entirely new hypotheses (e.g. regarding how differences in functional connectivity of different brain regions
between different cognitive tasks might be impacted by aging, task characteristics, or other factors).