Efficient learning algorithms for changing environments

By Elad Hazan, Seshadhri Comandur. 2009

219 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Management Science and Operations Research

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Management Science and Operations Research

Citations in the corpus

Adaptive Gradient-Based Meta-Learning Methods (2019, background): “The generality of ARUBA also allows for the incorporation of other dynamic regret bounds [25, 52] and other non-static notions of regret [27].”
Online Meta-Learning (2019, neutral): “To overcome these limitations, prior work has placed restrictions on how quickly the loss functions or the comparator model can change (Hazan & Comandur, 2009; Hall & Willett, 2015; Herbster & Warmuth, 1995).”
Efficient tracking of a growing number of experts (2017, background): “While our work belongs to the ﬁrst category, it should be possible to transport our regret bounds to the convex bounded case by using time-varying learning rates, as done e.g. by Hazan and Seshadhri (2009) and Gyorgy et al. (2012).”
Dynamic Regret of Strongly Adaptive Methods (2017, background): “To address this limitation, new forms of performance measure, including adaptive regret (Hazan & Seshadhri, 2007; 2009) and dynamic regret (Zinkevich, 2003; Hall & Willett, 2013), were proposed and received significant interest recently.”
Online Optimization in Dynamic Environments (2013, background): “been studied previously in the literature in the context of tracking regret (also known as shifting regret) [25, 26], and the closely-related concept of adaptive regret [21,27].”
Online Convex Optimization in Dynamic Environments (2015, background): “This idea is typically referred to as “tracking” or “shifting” regret [20], [21], which is closely-related to “adaptive” regret [22], [23].”
A Closer Look at Adaptive Regret (2012, methodology): “2 that it also underlies Follow the Leading History of [7].”
A new look at shifting regret (2012, methodology): “Hazan and Seshadhri(2009) essentially considered the case of online convex optimization with exp-concave loss function (see Section6 below).”
The Forget-me-not Process (2016, background): “Many variants have been proposed in the literature [20, 11, 10, 17], which trade off predictive accuracy for improved time and space complexity; in particular the Partition Tree Weighting meta-algorithm [17] has O(n log n) time and O(log n) space complexity, and has been shown empirically to exhibit”
Online Learning: A Comprehensive Survey (2018, methodology): “Popular approaches in this category include Aggregating Algorithms (Vovk and Watkins, 1998), Fast Universalization Algorithm (Akcoglu et al., 2004) and Follow the Leading History (Hazan and Seshadhri, 2009).”
Mirror Descent Meets Fixed Share (and feels no regret) (2012, unknown): “Adaptive regret was introduced by [8] and can be viewed as a variant of discounted regret where the monotonicity assumption is dropped.”
Dynamical Models and tracking regret in online convex programming (2013, background): “…or dynamic comparator sequence has been studied previously in the literature in the context of tracking regret, shifting regret (Herbster & Warmuth, 2001; Cesa-Bianchi et al., 2012), and the closely-related concept of adaptive regret (Littlestone & Warmuth, 1994; Hazan & Seshadhri, 2009).”
Shifting Regret, Mirror Descent, and Matrices (2016, methodology): “Note that the general wrapper algorithms derived for switching regret (see Hazan & Seshadhri, 2009; György et al., 2012; Adamskiy et al., 2012; Daniely et al., 2015) are not directly applicable to obtain shifting bounds.”