AdaGrad: Difference between revisions

Revision as of 17:24, 26 November 2021

Author: Daniel Villarraga (SYSEN 6800 Fall 2021)

Introduction

AdaGrad is a family of sub-gradient algorithms for stochastic optimization. The algorithms belonging to that family are similar to second-order stochastic gradient descend with an approximation for the Hessian of the optimized function. AdaGrad's name comes from Adaptative Gradient. Intuitively, it adapts the learning rate for each feature depending on the estimated geometry of the function; additionally, it tends to assign higher learning rates to infrequent features, which ensures that the parameter updates rely less on frequency and more on relevance.

AdaGrad was introduced by Duchi et al.^[1] in a highly cited paper published in the Journal of machine learning research in 2011. It is arguably one of the most popular algorithms for machine learning (particularly for training deep neural networks) and it influenced the development of the Adam algorithm^[2].

Theory

The objective of AdaGrad is to minimize the expected value of a stochastic objective function, with respect to a set of parameters, given a sequence of realizations of the function. As with other sub-gradient-based methods, it achieves so by updating the parameters in the opposite direction of the sub-gradients. While standard sub-gradient methods use update rules with step-sizes that ignore the information from the past observations, AdaGrad adapts the learning rates for each parameter individually using the sequence of gradient estimates.

Definitions

$f(x)$ : Stochastic objective function with parameters $x$ .

$f_{t}(x)$ : Realization of stochastic objective at time step $t$ . For simplicity $f_{t}$ .

$g_{t}(x)$ : The gradient of $f_{t}(x)$ with respect to $x$ , formally $\nabla _{x}f_{t}(x)$ . For simplicity, $g_{t}$ .

$x_{t}$ : Parameters at time step $t$ .

$G_{t}$ : Outer product of all previous subgradients, given by ${\textstyle \sum _{\tau =1}^{t}g_{\tau }g_{\tau }^{\top }}$

Standard Sub-gradient Update

Standard sub-gradient algorithms update parameters $x$ according to the following rule:

x_{t+1}=x_{t}-\eta g_{t}

where

\eta

denotes the step-size often refered as learning rate.

AdaGrad Update

Algorithm

Regret Bounds

Empirical Performance

Numerical Example

Applications

Summary and Discussion

References

↑ Duchi, J., Hazan, E., & Singer, Y. (2011). Adaptive subgradient methods for online learning and stochastic optimization. Journal of machine learning research, 12(7).
↑ Kingma, D. P., & Ba, J. (2014). Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980.

[1] Duchi, J., Hazan, E., & Singer, Y. (2011). Adaptive subgradient methods for online learning and stochastic optimization. Journal of machine learning research, 12(7).

[2] Kingma, D. P., & Ba, J. (2014). Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980.

[1]

[2]

@@ Line 7: / Line 7: @@
 == Theory ==
+The objective of AdaGrad is to minimize the expected value of a stochastic objective function, with respect to a set of parameters, given a sequence of realizations of the function. As with other sub-gradient-based methods, it achieves so by updating the parameters in the opposite direction of the sub-gradients. While standard sub-gradient methods use update rules with step-sizes that ignore the information from the past observations, AdaGrad adapts the learning rates for each parameter individually using the sequence of gradient estimates.
 === Definitions ===
@@ Line 20: / Line 21: @@
 <math>x_t</math>: Parameters at time step <math>t</math>.
-<math>G_t</math>: Outer product of all previous subgradients, given by <math>\sum_{\tau=1}^t g_{\tau}g_{\tau}^{\top}
+<math>G_t</math>: Outer product of all previous subgradients, given by <math display="inline">\sum_{\tau=1}^t g_{\tau}g_{\tau}^{\top}
 </math>
-=== Traditional Sub-gradient Update ===
+=== Standard Sub-gradient Update ===
+Standard sub-gradient algorithms update parameters <math>x</math> according to the following rule:
-=== Adagrad Update ===
+<math display="block">x_{t+1} = x_t - \eta g_t</math>where <math>\eta</math> denotes the step-size often refered as learning rate.
+=== AdaGrad Update ===
 === Algorithm ===