Research

Statistical Learning and Graphical Models

 

Christopher X. Ren, Sidhant Misra, Marc Vuffray, Andrey Y. Lokhov
Learning Continuous Exponential Families Beyond Gaussian
Submitted [ArXiv]

Short description: Introducing a new estimator ISODUS for continuous non-Gaussian exponential family distributions with unbounded support and multi-body interactions

 

 

Arkopal Dutt, Andrey Y. Lokhov, Marc Vuffray, Sidhant Misra,
Exponential Reduction in Sample Complexity with Learning of Ising Model Dynamics
International Conference on Machine Learning (ICML 2021) [ArXiv]

Short description: Learning of graphical models from correlated samples in the out-of-equilibrium regime is exponentially easier compared to the independent samples setting

  

 

Abhijith J., Andrey Y. Lokhov, Sidhant Misra, Marc Vuffray
Learning of discrete graphical models with neural networks
Advances in Neural Information Processing Systems (NeurIPS 2020) [ArXiv]

Short description: Discovering parsimonious basis representation with NeurISE, an Interaction Screening based estimator incorporating neural networks acting as universal energy function approximators

 

 

Marc Vuffray, Sidhant Misra, Andrey Y. Lokhov
Efficient learning of discrete graphical models
Advances in Neural Information Processing Systems (NeurIPS 2020) [ArXiv]

Short description: Learning discrete graphical models with arbitrary alphabets and multi-body interactions using GRISE, Generalized Regularized Interaction Screening Estimator

 

 

Sidhant Misra, Marc Vuffray, Andrey Y. Lokhov
Information theoretic optimal learning of Gaussian graphical models
Conference on Learning Theory (COLT 2020) [pdf] [ArXiv]

Short description: Beyond LASSO with SLICE and DICE algorithms that achieve the IT bound on sample complexity for learning the structure of Gaussian graphical models

Ising-1

Andrey Y. Lokhov, Marc Vuffray, Sidhant Misra, Michael Chertkov
Optimal structure and parameter learning of Ising models
Science Advances, 4 , e1700791 (2018) [pdf] [ArXiv] [Code]

Marc Vuffray, Sidhant Misra, Andrey Y. Lokhov, Michael Chertkov
Interaction screening: efficient and sample-optimal learning of Ising models
Advances in Neural Information Processing Systems (NIPS) 2016 [pdf] [ArXiv]

Short description: Sample-optimal “Interaction Screening” method for provably learning arbitrary binary graphical models without any assumptions

Andrey Y. Lokhov, Olga V. Valba, Mikhail V. Tamm, Sergei K. Nechaev
Phase transition in random planar diagrams and RNA-type matching
Phys. Rev. E 88, 052117 (2013) [pdf] [ArXiv]

Andrey Y. Lokhov, Olga V. Valba, Sergei K. Nechaev, Mikhail V. Tamm
Topological transition in disordered planar matching: combinatorial arcs expansion
J. Stat. Mech. P12004 (2014) [pdf] [ArXiv]

Short description: Combinatorics of RNA-type matching structures and new phase transition

Dynamic Message-Passing and Spreading Processes

Mateusz Wilinski, Lauren Castro, Jeffrey Keithley, Carrie Manore, Josefina Campos, Ethan Romero-Severson, Daryl Domman, Andrey Y. Lokhov
Congruity of genomic and epidemiological data in modeling of local cholera outbreaks
Submitted [ArXiv]

Short description: We use high-fidelity case count and whole genome sequencing data from the 1991-1998 cholera epidemic in Argentina, and show that consistency between the epidemiological model parameters estimated from both genetic and case-count data sources.

Mateusz Wilinski, Andrey Y. Lokhov
Scalable Learning of Independent Cascade Dynamics from Partial Observations
International Conference on Machine Learning (ICML 2021) [ArXiv]

Short description: Introducing a scalable algorithm SLICER that estimates parameters of the Independent Cascade model. In the context of learning for inference, tractable inference from the learned model generates a better prediction of marginal probabilities compared to the original model

 

Andrey Y. Lokhov, David Saad
Scalable Influence Estimation Without Sampling
Submitted [ArXiv]

Short description: Scalable dynamic message-passing algorithm for estimation of spread in the Independent Cascade type diffusion models

Hanlin Sun, David Saad, Andrey Y. Lokhov
Competition, Collaboration, and Optimization in Multiple Interacting Spreading Processes
Phys. Rev. X 11, 011048 (2021) [pdf] [ArXiv]

Short description: Exact dynamic message-passing equations for estimation of marginal infection probabilities for collaborative and mutually exclusive epidemics, and their use for the optimal resource allocation problem

 

Andrey Y. Lokhov, David Saad
Optimal deployment of resources for maximizing impact in spreading processes
Proceedings of the National Academy of Sciences, 114 (39) E8138-E8146 (2017) [pdf] [ArXiv]

Short description: Optimal targeting in spreading processes with dynamic message-passing equations and forward-backward optimization method used in artificial neural networks

Andrey Y. Lokhov
Reconstructing parameters of spreading models from partial observations
Advances in Neural Information Processing Systems (NIPS) 2016 [pdf] [ArXiv] [video]

Andrey Y. Lokhov, Theodor Misiakiewicz
Efficient reconstruction of transmission probabilities in a spreading process from partial observations
Work in progress [ArXiv]

Short description: Introducing a dynamic message-passing algorithm DMPrec for learning parameters of spreading models from partial observations 

 

Andrey Y. Lokhov, Marc Mézard, Lenka Zdeborová
Dynamic message-passing equations for models with unidirectional dynamics
Phys. Rev. E 91, 012811 (2015) [pdf] [ArXiv]

Short description: Solution of many dynamic models (random field Ising model, epidemic and rumor spreading, threshold models) on given network instances

 

Andrey Y. Lokhov, Marc Mézard, Hiroki Ohta, Lenka Zdeborová
Inferring the origin of an epidemic with a dynamic message-passing algorithm
Phys. Rev. E 90, 012801 (2014)  [pdf] [ArXiv]

Short description: Localization of the epidemic source from a partial snapshot at unknown time

Quantum Computing

Byron Tasseff et al.
On the Emerging Potential of Quantum Annealing Hardware for Combinatorial Optimization

Submitted [ArXiv]

Short description: We demonstrate the existence of classes of contrived optimization problems where D-Wave Systems’ most recent Advantage Performance Update quantum annealer provides run time benefits over a collection of established classical solution methods.

 

Zachary Morrell et al.
Signatures of Open and Noisy Quantum Systems in Single-Qubit Quantum Annealing
Submitted [ArXiv]

Short description: This paper shows that both thermal and magnetic field fluctuations are key sources of noise that need to be included in an open quantum system model to reproduce the output statistics of the hardware.

 

Abhijith J. et al.
Quantum Algorithm Implementations for Beginners

ACM Transactions on Quantum Computing, Volume 3, Issue 4, 18, pp. 1–92 (2022) [ArXiv]

Short description: An introduction to quantum computing algorithms and their implementation on IBM QX quantum computer

 

 

 

Adrien Suau, Marc Vuffray, Andrey Y. Lokhov, Lukasz Cincio, Carleton Coffrin
Vector Field Visualization of Single-Qubit State Tomography

Accepted to IEEE QCE 2022 [ArXiv]

Short description: Developing a vector field visualization for quantum state tomography characterization of individual qubits, and demonstration of qubit performance features in IBM quantum computing hardware.

 

Jon Nelson, Marc Vuffray, Andrey Y. Lokhov, Tameem Albash, and Carleton Coffrin
High-Quality Thermal Gibbs Sampling with Quantum Annealing Hardware
Phys. Rev. Applied 17, 044046 (2022) [ArXiv]

Short description: Introducing a procedure for producing high-quality samples from quantum annealing hardware

 

Marc Vuffray, Carleton Coffrin, Yaroslav Kharkov, Andrey Y. Lokhov
Programmable Quantum Annealers as Noisy Gibbs Samplers
PRX Quantum 3, 020317 (2022) [ArXiv]

Short description: Characterization of quantum annealers’ sampling properties using statistical learning methods, including unexpected spurious interactions in the output distribution due to qubit noise

Jon Nelson, Marc Vuffray, Andrey Y. Lokhov, Carleton Coffrin
Single-Qubit Fidelity Assessment of Quantum Annealing Hardware
IEEE Transactions on Quantum Engineering, 2, 1-10 (2021) [ArXiv]

Adrien Suau et al.
Single-Qubit Cross Platform Comparison of Quantum Computing Hardware
Submitted [ArXiv]

Short description: Quantifying the error performance of individual qubits in quantum annealing and gate quantum computers

 


The Potential of Quantum Annealing for Rapid Solution Structure Identification
Yuchen Pang, Carleton Coffrin, Andrey Y. Lokhov, Marc Vuffray
Constraints (2020) [ArXiv]
Presented at CPAIOR 2020

Short description: Identification of a hard instance of an optimization problem where quantum annealing provides notable performance gains over established classical algorithms

Dynamical Systems, Power Grid, and Cyber-Physical Systems

 

Zheguang Zhao, Deepjyoti Deka, Andrey Y. Lokhov
Learning of Cyber-Physical Systems
Work in progress

Short description: Learning of an effective cyber-physical model from discrete and continuous time series of physical and control processes

 

Jordan Snyder, Anatoly Zlotnik, Andrey Y. Lokhov
Data-driven Selection of Coarse-Grained Models of Coupled Oscillators
Phys. Rev. Research 2, 043402 (2020) [pdf] [ArXiv]

Short description: Learning of macroscopic reduced-order models in systems of coupled oscillators from coarse-grained microscopic data

 

 

Bo Li, David Saad, Andrey Y. Lokhov
Reducing Urban Traffic Congestion Due To Localized Routing Decisions
Phys. Rev. Research 2, 032059(R) (2020) [pdf] [ArXiv]

Short description: Discovery of paradoxical traffic patterns emerging within a new traffic model that includes localized routing inducement, and development of a scalable optimization algorithm for identifying mechanisms to minimize congestion

 

 

Christopher Hannon, Deepjyoti Deka, Dong Jin, Marc Vuffray, Andrey Y. Lokhov
Real-time Anomaly Detection and Classification in Streaming PMU Data
IEEE PowerTech 2021 [ArXiv]
Presented in a demo track of NeurIPS 2019

Short description: Anomaly detection and classification in streaming phasor measurement units data via real-time learning of effective dynamical model

Andrey Y. Lokhov, Marc Vuffray, Dmitry Shemetov, Deepjyoti Deka, Michael Chertkov
Online Learning of Power Transmission Dynamics
PSCC 2018 [ArXiv]Andrey Y. Lokhov, Deepjyoti Deka, Marc Vuffray, Michael Chertkov
Uncovering Power Transmission Dynamic Model from Incomplete PMU Observations
CDC 2018Deepjyoti Deka, Armin Zare,  Andrey Y. Lokhov, Mihailo Jovanovic, Michael Chertkov
Estimation of state and noise covariance in power grids using limited nodal PMUs
IEEE Global Conference on Signal and Information Processing (2017) [pdf]

 

Andrey Y. Lokhov, Nathan Lemons, Thomas C. McAndrew, Aric Hagberg, Scott Backhaus
Detection of cyber-physical faults and intrusions from physical correlations
IEEE 16th International Conference on Data Mining Workshops (ICDMW), 303-310 (2016)
[pdf] [ArXiv]
Presented at “Outlier Definition, Detection, and Description on Demand” workshop at KDD 2016

Short description: Detection of anomalies in cyber-physical systems from real data

Comments are closed.