arXiv subjects

Jihwan Lee

Publications and source records attributed to Jihwan Lee.

41 records · Page 3Linked to original sources

arXiv2021

Human Brain Mapping with Multi-Thousand Channel PtNRGrids Resolves Novel Spatiotemporal Dynamics

Electrophysiological devices are critical for mapping eloquent and diseased brain regions and for therapeutic neuromodulation in clinical settings and are extensively utilized for research in brain-machine interfaces. However, the existing devices are often limited in either spatial resolution or cortical coverage, even including those with thousands of channels used in animal experiments. Here, we developed scalable manufacturing processes and dense connectorization to achieve reconfigurable thin-film, multi-thousand channel neurophysiological recording grids using platinum-nanorods (PtNRGrids). With PtNRGrids, we have achieved a multi-thousand channel array of small (30 μm) contacts with low impedance, providing unparalleled spatial and temporal resolution over a large cortical area. We demonstrate that PtNRGrids can resolve sub-millimeter functional organization of the barrel cortex in anesthetized rats that captured the histochemically-demonstrated structure. In the clinical setting, PtNRGrids resolved fine, complex temporal dynamics from the cortical surface in an awake human patient performing grasping tasks. Additionally, the PtNRGrids identified the spatial spread and dynamics of epileptic discharges in a patient undergoing epilepsy surgery at 1 mm spatial resolution, including activity induced by direct electrical stimulation. Collectively, these findings demonstrate the power of the PtNRGrids to transform clinical mapping and research with brain-machine interfaces and highlights a path toward novel therapeutics.

physics.app-ph↗

arXiv2021

Ultra-Sharp Nanowire Arrays Natively Permeate, Record, and Stimulate Intracellular Activity in Neuronal and Cardiac Networks

Intracellular access with high spatiotemporal resolution can enhance our understanding of how neurons or cardiomyocytes regulate and orchestrate network activity, and how this activity can be affected with pharmacology or other interventional modalities. Nanoscale devices often employ electroporation to transiently permeate the cell membrane and record intracellular potentials, which tend to decrease rapidly to extracellular potential amplitudes with time. Here, we report innovative scalable, vertical, ultra-sharp nanowire arrays that are individually addressable to enable long-term, native recordings of intracellular potentials. We report large action potential amplitudes that are indicative of intracellular access from 3D tissue-like networks of neurons and cardiomyocytes across recording days and that do not decrease to extracellular amplitudes for the duration of the recording of several minutes. Our findings are validated with cross-sectional microscopy, pharmacology, and electrical interventions. Our experiments and simulations demonstrate that individual electrical addressability of nanowires is necessary for high-fidelity intracellular electrophysiological recordings. This study advances our understanding of and control over high-quality multi-channel intracellular recordings, and paves the way toward predictive, high-throughput, and low-cost electrophysiological drug screening platforms.

cond-mat.mtrl-sci↗

arXiv2019

Continuous Learning for Large-scale Personalized Domain Classification

Domain classification is the task of mapping spoken language utterances to one of the natural language understanding domains in intelligent personal digital assistants (IPDAs). This is a major component in mainstream IPDAs in industry. Apart from official domains, thousands of third-party domains are also created by external developers to enhance the capability of IPDAs. As more domains are developed rapidly, the question of how to continuously accommodate the new domains still remains challenging. Moreover, existing continual learning approaches do not address the problem of incorporating personalized information dynamically for better domain classification. In this paper, we propose CoNDA, a neural network based approach for domain classification that supports incremental learning of new classes. Empirical evaluation shows that CoNDA achieves high accuracy and outperforms baselines by a large margin on both incrementally added new domains and existing domains.

cs.LG↗

arXiv2019

Locale-agnostic Universal Domain Classification Model in Spoken Language Understanding

In this paper, we introduce an approach for leveraging available data across multiple locales sharing the same language to 1) improve domain classification model accuracy in Spoken Language Understanding and user experience even if new locales do not have sufficient data and 2) reduce the cost of scaling the domain classifier to a large number of locales. We propose a locale-agnostic universal domain classification model based on selective multi-task learning that learns a joint representation of an utterance over locales with different sets of domains and allows locales to share knowledge selectively depending on the domains. The experimental results demonstrate the effectiveness of our approach on domain classification task in the scenario of multiple locales with imbalanced data and disparate domain sets. The proposed approach outperforms other baselines models especially when classifying locale-specific domains and also low-resourced domains.

cs.LG↗

arXiv2016

Mining Statistically Significant Attribute Associations in Attributed Graphs

Recently, graphs have been widely used to represent many different kinds of real world data or observations such as social networks, protein-protein networks, road networks, and so on. In many cases, each node in a graph is associated with a set of its attributes and it is critical to not only consider the link structure of a graph but also use the attribute information to achieve more meaningful results in various graph mining tasks. Most previous works with attributed graphs take into ac- count attribute relationships only between individually connected nodes. However, it should be greatly valuable to find out which sets of attributes are associated with each other and whether they are statistically significant or not. Mining such significant associations, we can uncover novel relationships among the sets of attributes in the graph. We propose an algorithm that can find those attribute associations efficiently and effectively, and show experimental results that confirm the high applicability of the proposed algorithm.

cs.SI↗