A convergence theory for deep learning via over-parameterization
Deep neural networks (DNNs) have demonstrated dominating performance in many fields;
since AlexNet, networks used in practice are going wider and deeper. On the theoretical …
since AlexNet, networks used in practice are going wider and deeper. On the theoretical …
Non-asymptotic identification of lti systems from a single trajectory
We consider the problem of learning a realization for a linear time-invariant (LTI) dynamical
system from input/output data. Given a single input/output trajectory, we provide finite time …
system from input/output data. Given a single input/output trajectory, we provide finite time …
Improper learning for non-stochastic control
M Simchowitz, K Singh… - Conference on Learning …, 2020 - proceedings.mlr.press
We consider the problem of controlling a possibly unknown linear dynamical system with
adversarial perturbations, adversarially chosen convex loss functions, and partially …
adversarial perturbations, adversarially chosen convex loss functions, and partially …
The nonstochastic control problem
We consider the problem of controlling an unknown linear dynamical system in the presence
of (nonstochastic) adversarial perturbations and adversarial convex loss functions. In …
of (nonstochastic) adversarial perturbations and adversarial convex loss functions. In …
Provable reinforcement learning with a short-term memory
Real-world sequential decision making problems commonly involve partial observability,
which requires the agent to maintain a memory of history in order to infer the latent states …
which requires the agent to maintain a memory of history in order to infer the latent states …
Online linear quadratic control
We study the problem of controlling linear time-invariant systems with known noisy dynamics
and adversarially chosen quadratic losses. We present the first efficient online learning …
and adversarially chosen quadratic losses. We present the first efficient online learning …
Revisiting ho–kalman-based system identification: Robustness and finite-sample analysis
Weconsider the problem of learning a realization for a linear time-invariant (LTI) dynamical
system from input/output data. Given a single input/output trajectory, we provide finite time …
system from input/output data. Given a single input/output trajectory, we provide finite time …
Model-free linear quadratic control via reduction to expert prediction
Y Abbasi-Yadkori, N Lazic… - The 22nd International …, 2019 - proceedings.mlr.press
Abstract Model-free approaches for reinforcement learning (RL) and continuous control find
policies based only on past states and rewards, without fitting a model of the system …
policies based only on past states and rewards, without fitting a model of the system …
Finite-time identification of linear systems: Fundamental limits and optimal algorithms
Y Jedra, A Proutiere - IEEE Transactions on Automatic Control, 2022 - ieeexplore.ieee.org
We investigate the linear system identification problem in the so-called fixed budget and
fixed confidence settings. In the fixed budget setting, the learner aims at estimating the state …
fixed confidence settings. In the fixed budget setting, the learner aims at estimating the state …
Sample complexity lower bounds for linear system identification
Y Jedra, A Proutiere - 2019 IEEE 58th Conference on Decision …, 2019 - ieeexplore.ieee.org
This paper establishes problem-specific sample complexity lower bounds for linear system
identification problems. The sample complexity is defined in the PAC framework: it …
identification problems. The sample complexity is defined in the PAC framework: it …