Current research descriptionI am currently a research scientists at Amazon Alexa in Cambridge, MA. I work on speech recognition and deep learning for wake word detection at Alexa. Below are my past areas of interest in academia.
Intent detection in EEG data and EEG-based brain-computer interfaces
Brain computer/machine interfaces is a communication paradigm that provides control of machines and communications by interfacing directly with human nervous system/brain. Brain computer interfaces have progressed rapidly over past years, propelled by successes in a variety of basic and applied fields of research ranging from machine learning to basic neuroscience. I focused on the development of methods and necessary computational tools for brain computer interfaces based on electroencephalographic (EEG) imaging of brain activity. This project aimed to realize high-performance control of a robotic manipulator via EEG and was supported by TUBITAK ARDEB 1001 grant No 113E611, the Science Academy (Bilim Akademesi, Turkey) BAGEP Young Scientist Award, and Toros University BAP grant No TUBAP135001.
Mishchenko Y et al. (2019) IEEE Transactions on Biomedical Engineering 66 (4), 977-987
Aci C et al. (2019) Expert Systems with Applications, 134, 153-166.
Kaya M et al. (2018) Scientific Data, 5, 180211
Mishchenko Y (2017) NeuroTalk 2017
Mishchenko Y at al. (2016) SFN 2016
Ozbay E et al. (2016) IMSEC 2016
Mishchenko Y (2016) SIU 2016
Mishchenko Y et al. (2016) "A Brain-Computer Interface detection of right and left hand movement imageries from EEG data using the SVM machine learning method" (preprint)
Yanar H, Mishchenko Y (2016) SIU 2016
Kaya M et al (2016) SIU 2016
Mishchenko Y, Kaya M (2015) SIU 2015
Methods for empirical characterization of detailed structure of the brain's neural circuits
I am interested in large-scale/high-throughput methods for empirically elucidating detailed structure and functions of neuronal circuits in the brain, by means of anatomical and functional imaging approaches.
Mishchenko Y (2016) "Recent Advances in Neural Connectivity Inference Problem for Very Large Scale Population Calcium Imaging." (book chapter) Neuroimaging, SM E-books
Mishchenko Y (2016) Journal of Computational Neuroscience, 41, 158
Mishchenko Y (2015) "Reconstructing functional neural circuits with single cell resolution: statistical methods for inferring neural network topology from large scale neural activity imaging data" (talk) Janelia Research Campus
Marblestone A, et al. (2014) arXiv:1404.5103
Rah J-C, et al. (2013) Frontiers in Neural Circuits, 7, 177
Marblestone A, et al. (2013) BioRxiv 001214
Mishchenko Y, Paninski L (2012) Journal of Computational Neuroscience, 33, 371
Mishchenko Y (2011) Journal of Neuroscience Methods 196, 289
Mishchenko Y, Vogelstein J, Paninski L (2011) Annals of Applied Statistics, 5, 1229
Mishchenko Y, Paninski L (2011) Annals of Applied Statistics, 5, 1893
NETFIT package - Fitting neural networks to population calcium imaging data, used in the above (Mishchenko, Vogelstein, and Paninski, AOAS 2011).
River-Alba M, et al. (2011) Current Biology, 21, 2000
Mishchenko Y, et al. (2010) Neuron, 67, 1009
Mishchenko Y (2010) PLoS ONE 5: e8853
Mishchenko Y (2009) J of Neurosci Methods 176, 276
Intelligent communication systems
I am interested in "intelligent" paradigms for communications in highly connected, large online communities. One of those projects - Evion - was a "smart" message routing technology for Twitter, similar to later Twitter Suggestions, Google Priority Inbox, Facebook Smart Feed, etc. Evion was concerned with discovering and serving in real-time personalized feeds of relevant tweets to project subscribers, connecting Twitterers with interesting tweets that otherwise remained outside of their usual Twitter feed. The project broadcasted from a series of Twitter accounts (see the Evion website below) via @-mentions, and used machine learning with crowd filtering to select individualized messages. The project was online from August 2011 till May 2012. Due to technical difficulties with Twitter, the project shut down in May 2012, having gathered over 6500 followers. MyScienceHighlights is another project of this type. This project aimed to help academics stay in touch with recent developments in periodic research literature in their respective research areas. The project aggregated Table of Contents from over 2000 journals in 200 academic areas, and selects personalized article feeds for its subscribers. MyScienceHighlights was online since February 2011 until 2018, where I shut down this service after having joined Amazon, amidst new offerings of similar type from Google scholar and Mendeley.
Click here to go to homepage.