Online Learning and Online Convex Optimization

By Shai Shalev-Shwartz. 2012

2,495 citations recorded by the source corpus.

Subfields: Computer Networks and Communications, Management Science and Operations Research, Optimization & Training

How researchers see this paper

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

In Computer Networks and Communications

In Management Science and Operations Research

In Optimization & Training

Citations in the corpus

Trust Region Policy Optimization (2015, methodology): “Relatedly, the policy update is the same update that is performed in online mirror descent [14], using entropy as the regularizer.”
Memory-Efficient Adaptive Optimization for Large-Scale Learning (2019, background): “We henceforth assume the general online optimization setting (see [22, 13]).”
Practical and Private (Deep) Learning without Sampling or Shuffling (2021, methodology): “The arguments extend to non-differentiable convex losses via standard use of sub-differentials [32, 64].”
Central Server Free Federated Learning over Single-sided Trust Social Networks (2019, methodology): “Shalev-Shwartz, S. et al. (2012). Online learning and online convex optimization. Foundations and Trends R © in Machine Learning, 4(2):107–194. Shalev-Shwartz, S. and Zhang, T. (2013). Stochastic dual coordinate ascent methods for regularized loss minimization.”
Adaptive Gradient-Based Meta-Learning Methods (2019, background): “FTL, over BR(θ t ||·); this is well-known to be logarithmic for the case R(·) = 12‖ · ‖ 2 2 (Shalev-Shwartz, 2011) and can be shown to be so in general by a novel strongly-convex coupling argument.”
Differentially Private Learning with Adaptive Clipping (2019, positive): “Because the loss is convex and has gradients bounded by 1, we can produce an online estimate of C that converges to the γ th quantile of X using online gradient descent (see, e.g., Shalev-Shwartz (2012)).”
Semi-Cyclic Stochastic Gradient Descent (2019, background): “See, e.g., Shalev-Shwartz (2012), for a development of SGD without the projection as in (2), and proof of the regret guarantee (16) for these updates.”
Decentralized Online Learning: Take Benefits from Others’ Data without Sharing Your Own to Track Global Trend (2019, methodology): “The reason is that the regret of an online algorithm is O (√T ) for convex loss functions [12, 31].”
Differentially Private Fair Learning (2018, methodology): “Thus, by Corollary 2.14 of [Shalev-Shwartz, 2012], we have that with probability at least 1 ´ β { 2, for any λ 1 P Λ 1 , The following two lemmas are taken from [Agarwal et al., 2018] and are used in the proof of Theorem 4.6 and Theorem 4.8.”
Certified Defenses for Data Poisoning Attacks (2017, methodology): “There are many algorithms that have this property (Shalev-Shwartz, 2011); the particular algorithm depicted in Algorithm 1 is a variant of regularized dual averaging (Xiao, 2010).”
How To Make the Gradients Small Stochastically: Even Faster Convex and Nonconvex SGD (2018, background): “The following definition and properties of Fenchel dual for convex functions is classical, and can be found for instance in the textbook [26].”
Asynchronous Parallel Stochastic Gradient for Nonconvex Optimization (2015, background): “same convergence rate [Xiao, 2009]. The stochastic gradient is also closely related to online learning algorithms. Please refer to the online learning literatures, for example, [Crammer et al., 2006, Shalev-Shwartz, 2011, Yang et al., 2014]. Most studies for stochastic gradient focus on convex optim”
Online Continual Learning with Natural Distribution Shifts: An Empirical Study with Visual Data (2021, background): “Following the common definition of online learning [11, 28], we define OCL as a game between a learner and an environment.”
Continual Lifelong Learning in Natural Language Processing: A Survey (2020, background): “+ on-line learning + forward transfer – no backward transfer – no knowledge retention – single task/domain (Bottou, 1999) (Bottou and LeCun, 2004) (Cesa-Bianchi and Lugosi, 2006) (Shalev-Shwartz, 2012) (C.”
Continual Prototype Evolution: Learning Online from Non-Stationary Data Streams (2020, background): “However, online learning typically makes the rigorous assumption of learning from an iid data stream [41] or does not focus on preventing catastrophic forgetting [34, 26], whereas data incremental learning disregards the iid assumption and additionally generalizes beyond the online learning horizon ”
Machine Unlearning (2019, background): “Morever, data pipelines are often not static [3]: new data is collected regularly and incrementally used to further reﬁne existing models following the online learning paradigm [4].”
Task-Free Continual Learning (2018, methodology): “We refer to [5, 33] for surveys and overviews on the topic.”
Agile Autonomous Driving using End-to-End Deep Imitation Learning (2017, background): “tive function for the online learning [27] approach to IL.”