Panagiotis Lymperopoulos

Panagiotis Lymperopoulos

PhD Candidate at Tufts University Computer Science

Publications

A neurosymbolic cognitive architecture framework for handling novelties in open worlds

Published in Artificial Intelligence Journal, 2024

We propose a neurosymbolic cognitive architecture framework for handling novelties in open worlds.

Recommended citation: Goel, S., Lymperopoulos, P., Thielstrom, R., Krause, E., Feeney, P., Lorang, P., Schneider, S., Wei, Y., Kildebeck, E., Goss, S., & others (2024). A neurosymbolic cognitive architecture framework for handling novelties in open worlds. Artificial Intelligence, 331, 104111. https://www.sciencedirect.com/science/article/pii/S0004370224000866

Graph Pruning for Enumeration of Minimal Unsatisfiable Subsets

Published in International Conference on Artificial Intelligence and Statistics, 2024

We propose a novel approach to enumerate minimal unsatisfiable subsets by pruning the search space using graph-based techniques.

Recommended citation: Lymperopoulos, P., & Liu, L. (2024). Graph Pruning for Enumeration of Minimal Unsatisfiable Subsets. International Conference on Artificial Intelligence and Statistics, 2647-2655. https://proceedings.mlr.press/v130/lymperopoulos21a.html

Oh, Now I See What You Want: Learning Agent Models with Internal States from Observations

Published in AAMAS2024, 2024

We propose a novel approach to learn agent models with internal states from observations.

Recommended citation: Lymperopoulos, P., & Scheutz, M. (2024). Oh, Now I See What You Want: Learning Agent Models with Internal States from Observations. In Proceedings of the 23rd International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2024). https://www.ifaamas.org/Proceedings/aamas2024/pdfs/p1.pdf

Exploiting variable correlation with masked modeling for anomaly detection in time series.

Published in Spotlight Presentation at RobustSeq @ NeurIPS 2022., 2022

Online anomaly detection in time series is a challenging task, especially when the time-series are stochastic. We propose a novel approach to exploit the correlation between variables in time series data to improve the performance of anomaly detection.

Recommended citation: Lymperopoulos, P., Li, Y., & Liu, L. (2022). Exploiting variable correlation with masked modeling for anomaly detection in time series. https://openreview.net/pdf?id=TCJuzs585W

Branching principles of animal and plant networks identified by combining extensive data, machine learning and modelling.

Published in Journal of the Royal Society Interface, 2021

Modelling the branching of vascular network tissue as asymmetric fractal structures. The model is further investigate in its ability to describe and discriminate between various species.

Recommended citation: Brummer, A. B., Lymperopoulos, P., Shen, J., Tekin, E., Bentley, L. P., Buzzard, V., Gray, A., Oliveras, I., Enquist, B. J., & Savage, V. M. (2021). Branching principles of animal and plant networks identified by combining extensive data, machine learning and modelling. Journal of the Royal Society Interface, 18 (174), 20200624. https://royalsocietypublishing.org/doi/10.1098/rsif.2020.0624

Integrating planning, execution and monitoring in the presence of open world novelties: Case study of an open world monopoly solver.

Published in arxiv preprint, 2021

Monopoly-playing agent that operates in an Open world, where game rules, elements and concepts are subject to change at test time. Gopalakrishnan, Sriram, et al. "Integrating Planning, Execution and Monitoring in the presence of Open World Novelties: Case Study of an Open World Monopoly Solver."

Recommended citation: arXiv preprint arXiv:2107.04303 (2021). https://arxiv.org/abs/2107.04303

Forecasting covid-19 counts at a single hospital: A hierarchical bayesian approach.

Published in Poster in ICLR 2021 Workshop on Machine Learning for Preventing and Combating Pandemics, 2021

A bayesian model for forecasting the daily number of hospitalized COVID-19 patients at a single hospital site.

Recommended citation: Lee, A. H., Lymperopoulos, P., Cohen, J. T., Wong, J. B., & Hughes, M. C. (2021). Forecasting covid-19 counts at a single hospital: A hierarchical bayesian approach.Poster in ICLR 2021 Workshop on Machine Learning for Preventing and Combating Pandemics, arXiv preprint arXiv:2104.09327. https://arxiv.org/abs/2104.09327

Deep-learning-based image restoration of depth-resolved, label-free, two-photon images for the quantitative morphological and functional characterization of human cervical tissues.

Published in Imaging, Manipulation, and Analysis of Biomolecules, Cells, and Tissues XIX, 2021

Using denoising autoencoders to accelerate lable-free medical imaging.

Recommended citation: Polleys, C. M., Lymperopoulos, P., Thieu, H.-T., Genega, E., Liu, L., & Georgakoudi, I. (2021). Deep-learning-based image restoration of depth-resolved, label-free, two-photon images for the quantitative morphological and functional characterization of human cervical tissues. Imaging, Manipulation, and Analysis of Biomolecules, Cells, and Tissues XIX, 11647, 116470Z. https://spie.org/Publications/Proceedings/Paper/10.1117/12.2578650

Concept wikification for COVID-19

Published in Proceedings of the 1st Workshop on NLP for COVID-19 (Part 2) at EMNLP 2020., 2020

Wikification of COVID-19 texts using large language models trained on scientific papers.

Recommended citation: Lymperopoulos, P., Qiu, H., & Min, B. (2020). Concept wikification for COVID-19. Proceedings of the 1st Workshop on NLP for COVID-19 (Part 2) at EMNLP 2020. https://www.aclweb.org/anthology/2020.nlpcovid19-2.29 https://www.aclweb.org/anthology/2020.nlpcovid19-2.29