Nikos Kargas

pic

I am an Applied Scientist at Amazon Alexa AI, working on Natural Language Processing (NLP) techniques for improving expressivity and flexibility of Text-to-Speech (TTS) systems.

I received my PhD degree in Electrical and Computer Engineering from the University of Minnesota under the supervision of Professor Nikolaos D. Sidiropoulos in Dec. 2020. My PhD work leverages tensor decompositions for developing efficient, robust and interpretable ML models. Applications include non-parametric density estimation, missing data imputation, recommender systems, spatio-temporal data analysis and time-series forecasting.

News

Research

2022

Low-rank Characteristic Tensor Density Estimation Part I: Foundations
M. Amiridi, N. Kargas, N. D. Sidiropoulos
IEEE Transactions on Signal Processing, 2022
[arxiv]
Low-rank Characteristic Tensor Density Estimation Part II: Compression and latent density estimation
M. Amiridi, N. Kargas, N. D. Sidiropoulos
IEEE Transactions on Signal Processing, 2022
[arxiv]

2021

Information-Theoretic Feature Selection via Tensor Decomposition and Submodularity
M. Amiridi, N. Kargas, N. D. Sidiropoulos
IEEE Transactions on Signal Processing, 2021
[arxiv] [slides]
Multi-version Tensor Completion for Time-delayed Spatio-temporal Data
C. Qian, N. Kargas, C. Xiao, L. M. Glass, N. D. Sidiropoulos, J. Sun
International Joint Conference on Artificial Intelligence (IJCAI), 2021
(acceptance rate = 13.9%).
[arxiv] [slides] [poster] [bib]
@inproceedings{QianKar2021, title = {Multi-version Tensor Completion for Time-delayed Spatio-temporal Data}, author = {Qian, Cheng and Kargas, Nikos and Xiao, Cao and Glass, Lucas M. and Sidiropoulos, Nicholas D. and Sun, Jimeng}, booktitle = {Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, {IJCAI-21}}, pages = {2906--2912}, year = {2021} }
STELAR: Spatio-temporal Tensor Factorization with Latent Epidemiological Regularization
N. Kargas, C. Qian, N. D. Sidiropoulos, C. Xiao, L. M. Glass and J. Sun
AAAI Conference on Artificial Intelligence (AAAI), 2021
(acceptance rate = 21%).
Presentation (Presented by Professor Jimeng Sun)
[arxiv] [slides] [poster] [code] [bib]
@inproceedings{KarQian2021, title = {{STELAR}: Spatio-temporal Tensor Factorization with Latent Epidemiological Regularization}, booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence}, author = {Kargas, Nikos and Qian, Cheng and Sidiropoulos, Nicholas D. and Xiao, Cao and Glass, Lucas M. and Sun, Jimeng}, year = {2021}, pages = {4830-4837} }
Supervised Learning and Canonical Decomposition of Multivariate Functions
N. Kargas and N. D. Sidiropoulos
IEEE Transactions on Signal Processing, 2021
Presentation (Presented by Professor N. D. Sidiropoulos)
[bib]
@article{karSid2021, title = {Supervised learning and canonical decomposition of multivariate functions}, author = {Kargas, Nikos and Sidiropoulos, Nicholas D}, journal = {IEEE Transactions on Signal Processing}, volume = {69}, pages = {1097-1107}, year = {2021} }

2020

Supervised Learning via Ensemble Tensor Completion
N. Kargas and N. D. Sidiropoulos
Asilomar Conference on Signals, Systems and Computers (ASILOMAR), 2020
Presentation
[slides] [bib]
@inproceedings{KarSid2020b, author = {Kargas, Nikos and Sidiropoulos, Nicholas D.}, booktitle = {54th Asilomar Conference on Signals, Systems, and Computers}, title = {Supervised Learning via Ensemble Tensor Completion}, year = {2020}, pages = {196-199} }
Nonlinear System Identification via Tensor Completion
N. Kargas and N. D. Sidiropoulos
AAAI Conference on Artificial Intelligence (AAAI), 2020 (Spotlight)
(acceptance rate = 20.6%).
Presentation (Presented by Professor N. D. Sidiropoulos)
[arxiv] [slides] [poster] [code] [bib]
@inproceedings{KarSid2020a, title = {Nonlinear System Identification via Tensor Completion}, booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence}, author = {Kargas, Nikos and Sidiropoulos, Nicholas D.}, year = {2020}, pages = {4420-4427} }

2019

Low-Rank Tensor Models for Improved Multi-Dimensional MRI: Application to Dynamic Cardiac T1 Mapping
B. Yaman, S. Weingartner, N. Kargas, N. D. Sidiropoulos and M. Akcakaya
IEEE Transactions on Computational Imaging, 2019
[bib]
@article{YamWein2020, author = {Yaman, Burhaneddin and Weingärtner, Sebastian and Kargas, Nikolaos and Sidiropoulos, Nicholas D. and Akçakaya, Mehmet}, journal = {IEEE Transactions on Computational Imaging}, title = {Low-Rank Tensor Models for Improved Multidimensional {MRI}: Application to Dynamic Cardiac {$T_1$} Mapping}, year = {2019}, volume = {6}, pages = {194-207} }
Statistical Learning Using Hierarchical Modeling of Probability Tensors
M. Amiridi, N. Kargas and N. D. Sidiropoulos
IEEE Data Science Workshop (DSW), 2019 (Best student paper award)
[slides] [bib]
@inproceedings{AmiKar2019, author = {Amiridi, Magda and Kargas, Nikos and Sidiropoulos, Nicholas D.}, booktitle = {2019 IEEE Data Science Workshop (DSW)}, title = {Statistical Learning Using Hierarchical Modeling of Probability Tensors}, year = {2019}, pages = {290-294} }
Crowdsourcing via Pairwise Co-occurrences: Identifiability and Algorithms
S. Ibrahim, X. Fu, N. Kargas and K. Huang
Advances in Neural Information Processing Systems (NeurIPS), 2019
(acceptance rate = 21%).
[arxiv] [slides] [code] [bib]
@inproceedings{IbrFu2019, title = {Crowdsourcing via Pairwise Co-occurrences: Identifiability and Algorithms}, author = {Ibrahim, Shahana and Fu, Xiao and Kargas, Nikolaos and Huang, Kejun}, booktitle = {Advances in Neural Information Processing Systems (NeurIPS)}, pages = {7847--7857}, volume = {32}, year = {2019} }
Learning Mixtures of Smooth Product Distributions: Identifiability and Algorithm
N. Kargas and N. D. Sidiropoulos
International Conference on Artificial Intelligence and Statistics (AISTATS), 2019
(acceptance rate = 30%).
[arxiv] [poster] [code] [bib]
@inproceedings{KarSid2019, title = {Learning Mixtures of Smooth Product Distributions: Identifiability and Algorithm}, author = {Kargas, Nikos and Sidiropoulos, Nicholas D}, booktitle = {Proceedings of the 22nd International Conference on Artificial Intelligence and Statistics (AISTATS)}, pages = {388--396}, year = {2019} }

2018

Tensors, Learning, and `Kolmogorov Extension' for Finite-alphabet Random Vectors
N. Kargas, N. D. Sidiropoulos and X. Fu
IEEE Transactions on Signal Processing, 2018
[arxiv] [slides] [bib]
@article{KarSid2018, author = {N. {Kargas} and N. D. {Sidiropoulos} and X. {Fu}}, journal = {IEEE Transactions on Signal Processing}, title = {Tensors, Learning, and ``{Kolmogorov} Extension'' for Finite-Alphabet Random Vectors}, volume = {66}, number = {18}, pages = {4854--4868}, year = {2018} }

2017

Locally Low-Rank tensor regularization for high-resolution quantitative dynamic MRI
B. Yaman, S. Weingartner, N. Kargas, N. D. Sidiropoulos and M. Akcakaya
IEEE 7th International Workshop on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP), 2017
[bib]
Low-Rank Tensor Regularization for Improved Dynamic Quantitative Magnetic Resonance Imaging
N. Kargas, S. Weingartner, N. D. Sidiropoulos and M. Akcakaya
Signal Processing with Adaptive Sparse Structured Representations Workshop (SPARS), 2017
[poster] [bib]
Completing a Joint PMF from Projections: a Low-rank Coupled Tensor Factorization Approach
N. Kargas and N. D. Sidiropoulos
Information Theory and Applications Workshop (ITA), 2017
[arxiv] [slides] [bib]

2015

Fully-Coherent Reader with Commodity SDR for Gen2 FM0 and Computational RFID
N. Kargas, F. Mavromatis and A. Bletsas
IEEE Wireless Communications Letters, 2015
[code] [bib]

2014

Channel Coding for Increased Range Bistatic Backscatter Radio: Experimental Results
P. N. Alevizos, N. Fasarakis-Hilliard, K. Tountas, N. Agadakos, N. Kargas and A. Bletsas
IEEE RFID Technology and Applications Conference (RFID-TA), 2014
[bib]