• Corpus ID: 53290098

A Local Regret in Nonconvex Online Learning

@article{Aydre2018ALR,
  title={A Local Regret in Nonconvex Online Learning},
  author={Serg{\"u}l Ayd{\"o}re and Lee H. Dicker and Dean P. Foster},
  journal={ArXiv},
  year={2018},
  volume={abs/1811.05095}
}
We consider an online learning process to forecast a sequence of outcomes for nonconvex models. A typical measure to evaluate online learning algorithms is regret but such standard definition of regret is intractable for nonconvex models even in offline settings. Hence, gradient based definition of regrets are common for both offline and online nonconvex problems. Recently, a notion of local gradient based regret was introduced. Inspired by the concept of calibration and a local gradient based… 

Distributed Online Non-convex Optimization with Composite Regret

TLDR
A novel composite regret with a new network regret-based metric to evaluate distributed online optimization algorithms and DINOCO is shown to achieve sublinear regret, believed to be the first regret bound for general distributed online non-convex learning.

Time-Smoothed Gradients for Online Forecasting

TLDR
Inspired by the local regret metric that was introduced previously, this work proposes to use time-smoothed gradients within SGD update and shows that such a simple approach is computationally efficient compared to the alternatives.

RNN-based Online Learning: An Efficient First-Order Optimization Algorithm with a Convergence Guarantee

TLDR
An efficient first-order training algorithm is introduced that theoretically guarantees to converge to the optimum network parameters and is truly online such that it does not make any assumption on the learning environment to guarantee convergence.

References

SHOWING 1-8 OF 8 REFERENCES

Efficient Regret Minimization in Non-Convex Games

TLDR
A natural notion of regret is defined which permits efficient optimization and generalizes offline guarantees for convergence to an approximate local optimum and gradient-based methods that achieve optimal regret are given.

Variance Reduction for Faster Non-Convex Optimization

TLDR
This work considers the fundamental problem in non-convex optimization of efficiently reaching a stationary point, and proposes a first-order minibatch stochastic method that converges with an $O(1/\varepsilon)$ rate, and is faster than full gradient descent by $\Omega(n^{1/3})$.

Online Convex Programming and Generalized Infinitesimal Gradient Ascent

TLDR
An algorithm for convex programming is introduced, and it is shown that it is really a generalization of infinitesimal gradient ascent, and the results here imply that generalized inf initesimalgradient ascent (GIGA) is universally consistent.

A Multi-Horizon Quantile Recurrent Forecaster

We propose a framework for general probabilistic multi-step time series regression. Specifically, we exploit the expressiveness and temporal nature of Recurrent Neural Networks, the nonparametric

DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networks

New Algorithms for Learning Incoherent and Overcomplete Dictionaries

TLDR
This paper presents a polynomial-time algorithm for learning overcomplete dictionaries; the only previously known algorithm with provable guarantees is the recent work of Spielman, Wang and Wright who gave an algorithm for the full-rank case.

A Spectral Algorithm for Learning Hidden Markov Models

Asymptotic calibration

  • Biometrika
  • 1998