Sergül Aydöre

Senior Applied Scientist at Amazon Web Services - Machine Learning Researcher

I am a senior applied scientist at Amazon Web Services focusing on privacy and security problems. Prior to AWS, I was an Assistant Professor at the department of Electrical and Computer Engineering of Stevens Institute of Technology. I received my PhD degree from the Signal and Image Processing Institute at the University of Southern California in 2014 under the supervision of Richard M Leahy . My PhD thesis was on Measuring Functional Connectivity of the Brain . After finishing my PhD, my research interests have changed towards machine learning algorithms. I was lucky to work with great researchers Gaël Varoquaux , Bertrand Thirion , Dean Foster, Aaron Roth, and Michael Kearns.

Research

As a researcher, I am interested in robustness and generalization properties of machine learning models. Below are brief highlights of my research interests.

Robustness to Bad Samples
Robustness to Concept Drift
Robustness to Overfitting

Publications

  • S. Rho, S. Aydore , S. Tang, M. Kearns, A. Roth, Y. Wang, S. Wu, C. Archambeau , "Differentially Private Gradient Boosting on Linear Learners for Tabular Data", TSRML Workshop at NeurIPS 2022.

  • G. Vietri, C. Archambeau, S. Aydore , W. Brown, M. Kearns, A. Roth, A. Siva, S. Tang, S. Wu, "Private Synthetic Data for Multi-task Learning and Marginal Queries", In Proceedings of the 36th Conference on Neural Information Processing Systems (NeurIPS 2022).

  • Y. Kaya, M. C. Zafar, S. Aydore , N. Rauschmayr, K. Kenthapadi, "Generating Distributional Adversarial Examples to Evade Statistical Detectors", In Proceedings of the 39th International Conference on Machine Learning (ICML 2022).

  • H. Akrami, A.A. Joshi, S. Aydore , R. M. Leahy, "Deep Quantile Regression for Uncertainty Estimation inUnsupervised and Supervised Lesion Detection", IPMI Special Issue, Machine Learning for Biomedical Imaging Journal (MELBA), 2021.

  • H. Akrami, A.A. Joshi, J. Li, S. Aydore , R. M. Leahy, "A Robust Variational Autoencoder using Beta Divergence", Journal of Knowledge-based Systems, 2021.

  • E. Erdemir, J. Bickford, L. Melis, S. Aydore, , ``Adversarial Robustness with Non-uniform Perturbations '', In Proceedings of the 35th Conference on Neural Information Processing Systems (NeurIPS 2021), accessible at arXIV: 2102.12002.

  • S. Aydore, W. Brown, M. Kearns, K. Kenthapadi, L. Melis, A.Roth, A. Siva [alphabetical order], ``Differentially Private Query Release Through Adaptive Projection'', In Proceedings of the 38th International Conference on Machine Learning (ICML 2021).

  • H. Akrami, A. A. Joshi, S. Aydore, R. M. Leahy, ``Quantile Regression for Uncertainty Estimation in VAEs with Applications to Brain Lesion Detection'', International Conference on Information Processing in Medical Imaging, 2021.

  • H. Akrami, S. Aydore, A. A. Joshi, R. M. Leahy, ``Robust Variational Autoencoder for Tabular Data with Beta Divergence'', Uncertainty and Robustness in Deep Learning Workshop, ICML 2020, accesible at arXIV: 2006.08204.

  • H. Akrami, A. A. Joshi, J. Li, S. Aydore , R. M. Leahy, ``Brain Lesion Detection Using Variational Autoencoder and Transfer Learning'', IEEE International Symposium on Biomedical Imaging (ISBI), Iowa City, Iowa, USA, 2020.

  • S. Aydore , T. Zhu, D. Foster, ``Dynamic Local Regret for Non-convex Online Forecasting'', In Proceedings of the 33rd Conference on Neural Information Processing Systems (NeurIPS 2019), December 2019, Vancouver, CANADA, accesible at arXIV: 1910.07927.

  • S. Aydore , B. Thirion, G. Varoquaux. ``Feature Grouping as a Stochastic Regularizer for High-Dimensional Structured Data'', In Proceedings of the 36th International Conference on Machine Learning (ICML 2019), June 2019, Long Beach, CA, USA, accesible at arXIV: 1807.11718.

  • B. Wen, S. Aydore , ``ROMark: A Robust Watermarking System Using Adversarial Training'', Workshop on Machine Learning with Guarantees, NeurIPS 2019, Vancouver, CANADA, 2019, accessible at arXiV: 1910.01221.

  • T. Zhu, S. Aydore , ``Time-Smoothed Gradients for Online Forecasting'', Time Series Workshop, ICML 2019, Long Beach, CA, USA, 2019, accessible at arXiV: 1905.08850.

  • S. Aydore , L. Dicker, D. Foster.``A local Regret in Nonconvex Online Learning'', Continual Learning Workshop, NeurIPS 2018, Montreal, Canada, 2018, accessible at arXiv: 1811.05095.

  • S. Aydore , D. Pantazis, R. M. Leahy, ``A Note on the Phase Locking Value and its Properties,'' NeuroImage, February, 2013.

  • S. Aydore , S. Ashrafulla, A. A. Joshi, R. M. Leahy, ``A Measure of Connectivity in the Presence of Crosstalk,'' Asilomar Conference on Signals, Systems and Computers, November 3-6, 2013, Pacific Grove, CA, USA.

  • S. Aydore , D. Pantazis, R. M. Leahy, ``Phase Synchrony in Multivariate Gaussian Data with Applications to Cortical Networks'', IEEE International Symposium on Biomedical Imaging (ISBI), Barcelona, Spain, 2012.

  • S. Aydore , I. Sen, M. Kivanc Mihcak, Y. P. Kahya, ``Classification of Respiratory Signals by Linear Analysis,'' IEEE Engineering in Medicine and Biology Society, Minneapolis, Minnesota, USA, 2009.

  • S. Aydore , I. Sen, M. K. Mihcak, Y. P. Kahya, ``Classification of Wheeze in Respiratory Sounds by Linear Discriminant Method,'' IEEE Signal and Image Communications Applications, Antalya, Turkey, 2009.

  • S. Aydore , M. K. Mihcak, A. Akin, R. K. Ciftci, ``On Temporal Connectivity of PFC via Gauss-Markov Modeling of fNIRS Signals,'' IEEE Transactions on Biomedical Engineering, 2010.

PRE-PRINTS

  • L. Chen, P. Gautier, S. Aydore , ``DropCluster: A Structured Dropout for Convolutional Neural Networks'', Invited Talk at Women in Machine Learning Workshop, NeurIPS 2019, December 2019, Vancouver, CANADA, accessible at arXiv:2002.02997.

INDUSTRY

  • S. Aydore , L. Melis, B. Coskun,``RVAE-MMD: Robust Variational Autoencoder with Beta and MMD divergences'', Amazon Machine Learning Conference, October, 2020.

  • S. Aydore , L. Dicker, K. Torkkola, D. Foster, ``Online Learning for a Non-convex Algorithm at Forecasting'', Amazon Machine Learning Conference, April, 2018.

  • S. Aydore , L. Razoumov, A. Mahani, ``Speeding up Gamma Fitting in Forecasting Systems'', Amazon Machine Learning Conference, May, 2017.

PATENTS

  • S. Aydore , B. Coskun, L. Melis, ``Detecting Anomalous Events from Categorical Data Using Autoencoders'', filed by Amazon, US Patent Number: 11537902, granted in 2023.

  • S. Aydore , W. Brown, M. Kearns, K. Kenthapadi, L. Melis, A. Siva, A. Roth, ``Generating Relaxed Synthetic Data using Adaptive Projection'', filed by Amazon, US Patent Number: 11487765, granted in 2022.

  • S. Upadhyay, S. Aydore , T. Movva, ``Systems and methods for machine-learning augmented application monitoring'', filed by JP Morgan, US Patent number:11151473, granted in 2021.

  • Y. Kaya, M. B. Zafar, N. Rauschmayr, S. Aydore , K. Kenthapadi, ``Detecting suspicious user input distributions in deep neural networks'', filed by Amazon, December, 2021.

  • B. Can, S. Aydore , B. Coskun, W. Ding, Q. Cui, O. Y. Polyakov, ``Multi-cluster based semi-supervised anomaly detection'', filed by Amazon, December, 2021.

  • E. Erdemir, S. Aydore , M. Torkamani, J. Bickford, L. Melis, ``Adversarial Training of Deep Neural Networks Using Non-uniform Perturbations'', filed by Amazon, April, 2021.

MEDIA COVERAGE and blog POSTS

Awards

  • AWS Machine Learning Research Award to support my research at Stevens, 2019.

  • ICML Diversity and Inclusion Award, 2019.

  • Google Cloud Platform Research Credits, 2019.

  • Ming Hsieh Institute Ph.D. Scholar (given to six outstanding senior Ph.D. students in Electrical Engineering Department at USC), 2013.

  • Best PhD Poster Award, The International Conference on Biomagnetism (BIOMAG), 2012.

  • National Publication Award from Brain Research Society in Turkey, 2010.

  • Opportunity Grant from EducationUSA funded by the Bureau of Educational and Cultural Affairs of the US Department of State, 2009.

  • Viterbi School of Engineering Doctoral Fellowship Award at University of Southern California, 2009.

  • Necmi Tanyolac Award from Biomedical Engineering Institute at Bogazici University (for contributions to biomedical engineering), 2009.

  • Graduate scholarship from TUBITAK (The Scientific and Technological Council of Turkey) (2007-2009).

Teaching

  • Deep Learning taught in Spring 2019, Fall 2020 at the department of Electrical and Computer Engineering, Stevens Institute of Technology

Talks

  • Columbia Center of Artificial Technology Symposium, New York, NY, October, 2022.

  • Microsoft Research, New York, NY, October, 2019.

  • Delft Mechanical Engineering Talent Event, TU Delft, The Netherlands, October, 2019.

  • Amazon Web Services Science Meetings, New York, NY, September, 2019.

  • Ming Hsieh Institute Ph.D. Scholar (given to six outstanding senior Ph.D. students in Electrical Engineering Department at USC), 2013.

  • Department of Electrical and Computer Engineering, University of Southern California, Los Angeles, CA, June, 2019

  • Montreal Women in Machine Learning and Data Science meetup, Montreal, CA, December, 2018

  • Invited Talk at Women in Machine Learning at NeurIPS, Montreal, CA, February, 2019

  • Bogazici University, Istanbul, Turkey, December, 2018

  • Sabanci University, Istanbul, Turkey, December, 2018

  • Acibadem University, Istanbul, Turkey, December, 2018

  • S2A Seminar at Telecom ParisTech, Paris, France, June 2018.

  • Neural Networks for Forecasting Demand, Deep Learning Summit, Boston, MA, May, 2018.

  • Organization of Human Brain Mapping, Annual Meeting, (OHBM 2014), Hamburg, Germany, May, 2014. ( 5% oral talk rate)

  • Cognitive Neurophysiology Laboratory, University of California Los Angeles, Los Angeles, CA, USA, May, 2014.

  • Laboratory for Intelligent Imaging and Neural Computing, Columbia University, New York, NY, USA, May 2014.

  • 18th International Conference on Biomagnetism (BIOMAG 2012), Paris, France, August 2012.

Service

  • Co-organizer of workshop Synthetic Data for ML at NeurIPS 2022.

  • General Chair of Women in Machine Learning Workshop at NeurIPS 2022.

  • Co-leader of the breakout session "Does your model know what it doesn't know? Uncertainty Estimation and OOD Detection in DL" at WiML Un-workshop at ICML 2021.

  • Co-organizer of workshop Synthetic Data Generation: Quality, Privacy, Bias at ICLR 2021.

  • Co-leader with H. Akrami of the breakout session "Robust Machine Learning with Bad Training Data" at WiML Un-workshop at ICML 2020.

  • Area Chair for WiML workshop at NeurIPS 2019, NeurIPS 2020.

  • Reviewer for conferences NeurIPS ICML, ICLR, AISTATS

  • Contributions to open source machine learning Python Library scikit-learn

  • Reviewer for journals NeuroImage, PLOSOne, Mathematical Problems in Engineering

  • Member of Technical Program Committee for GlobalSIP Symposium on Big Data Analysis and Challenges in Medical Imaging, 2016.