Carlos Alberto Gomez-Uribe
Carlos Alberto Gomez-Uribe has a total of 3 co-authors. Here are the top co-authors by number of shared publications:
- Abraham Rami Tzafriri
Harvard-MIT Division of Health Sciences and Technology, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA
1 shared publication - Brian Karrer
Facebook Menlo Park California USA
1 shared publication - George C. Verghese
Laboratory for Electromagnetic and Electronic Systems Massachusetts Institute of Technology 02139 Cambridge MA
1 shared publication
Latest Publications
- 24 January 2026
- Published by Springer Nature in Quantum Studies: Mathematics and Foundations
- Vol. 13 (1) , 7
Planck’s law from a classical free energy extremum involving fisher information
- 24 January 2026
- Published by Springer Nature in Quantum Studies: Mathematics and Foundations
- Vol. 13 (1) , 7
Planck’s law from a classical free energy extremum involving fisher information
- Open Access
The decoupled extended Kalman filter for dynamic exponential-family factorization models
Motivated by the needs of online large-scale recommender systems, we specialize the decoupled extended Kalman filter (DEKF) to factorization models, including factorization machines, matrix and tensor factorization, and illustrate the effectiveness of the approach through numerical experiments on synthetic and on real-world data. Online learning of model parameters through the DEKF makes factorization models more broadly useful by (i) allowing for more flexible observations through the entire exponential family, (ii) modeling parameter drift, and (iii) producing parameter uncertainty estimates that can enable explore/exploit and other applications. We use a different parameter dynamics than the standard DEKF, allowing parameter drift while encouraging reasonable values. We also present an alternate derivation of the extended Kalman filter and DEKF that highlights the role of the Fisher information matrix in the EKF.
Open AccessThe decoupled extended Kalman filter for dynamic exponential-family factorization models
Motivated by the needs of online large-scale recommender systems, we specialize the decoupled extended Kalman filter (DEKF) to factorization models, including factorization machines, matrix and tensor factorization, and illustrate the effectiveness of the approach through numerical experiments on synthetic and on real-world data. Online learning of model parameters through the DEKF makes factorization models more broadly useful by (i) allowing for more flexible observations through the entire exponential family, (ii) modeling parameter drift, and (iii) producing parameter uncertainty estimates that can enable explore/exploit and other applications. We use a different parameter dynamics than the standard DEKF, allowing parameter drift while encouraging reasonable values. We also present an alternate derivation of the extended Kalman filter and DEKF that highlights the role of the Fisher information matrix in the EKF.
- Open Access
- 28 December 2008
- Published by AIP Publishing in The Journal of Chemical Physics
- Vol. 129 (24) , 244112
Enhanced identification and exploitation of time scales for model reduction in stochastic chemical kinetics
Widely different time scales are common in systems of chemical reactions and can be exploited to obtain reduced models applicable to the time scales of interest. These reduced models enable more efficient computation and simplify analysis. A classic example is the irreversible enzymatic reaction, for which separation of time scales in a deterministic mass action kinetics model results in approximate rate laws for the slow dynamics, such as that of Michaelis–Menten. Recently, several methods have been developed for separation of slow and fast time scales in chemical master equation (CME) descriptions of stochastic chemical kinetics, yielding separate reduced CMEs for the slow variables and the fast variables. The paper begins by systematizing the preliminary step of identifying slow and fast variables in a chemical system from a specification of the slow and fast reactions in the system. The authors then present an enhanced time-scale-separation method that can extend the validity and improve the accuracy of existing methods by better accounting for slow reactions when equilibrating the fast subsystem. The resulting method is particularly accurate in systems such as enzymatic and protein interaction networks, where the rates of the slow reactions that modify the slow variables are not a function of the slow variables. The authors apply their methodology to the case of an irreversible enzymatic reaction and show that the resulting improvements in accuracy and validity are analogous to those obtained in the deterministic case by using the total quasi-steady-state approximation rather than the classical Michaelis–Menten. The other main contribution of this paper is to show how mass fluctuation kinetics models, which give approximate evolution equations for the means, variances, and covariances of the concentrations in a chemical system, can feed into time-scale-separation methods at a variety of stages.
- 28 December 2008
- Published by AIP Publishing in The Journal of Chemical Physics
- Vol. 129 (24) , 244112
Open AccessEnhanced identification and exploitation of time scales for model reduction in stochastic chemical kinetics
Widely different time scales are common in systems of chemical reactions and can be exploited to obtain reduced models applicable to the time scales of interest. These reduced models enable more efficient computation and simplify analysis. A classic example is the irreversible enzymatic reaction, for which separation of time scales in a deterministic mass action kinetics model results in approximate rate laws for the slow dynamics, such as that of Michaelis–Menten. Recently, several methods have been developed for separation of slow and fast time scales in chemical master equation (CME) descriptions of stochastic chemical kinetics, yielding separate reduced CMEs for the slow variables and the fast variables. The paper begins by systematizing the preliminary step of identifying slow and fast variables in a chemical system from a specification of the slow and fast reactions in the system. The authors then present an enhanced time-scale-separation method that can extend the validity and improve the accuracy of existing methods by better accounting for slow reactions when equilibrating the fast subsystem. The resulting method is particularly accurate in systems such as enzymatic and protein interaction networks, where the rates of the slow reactions that modify the slow variables are not a function of the slow variables. The authors apply their methodology to the case of an irreversible enzymatic reaction and show that the resulting improvements in accuracy and validity are analogous to those obtained in the deterministic case by using the total quasi-steady-state approximation rather than the classical Michaelis–Menten. The other main contribution of this paper is to show how mass fluctuation kinetics models, which give approximate evolution equations for the means, variances, and covariances of the concentrations in a chemical system, can feed into time-scale-separation methods at a variety of stages.