ADASS: Adaptive Sample Selection for Training Acceleration

06/11/2019
by   Shen-Yi Zhao, et al.
Nanjing University
0

Stochastic gradient decent (SGD) and its variants, including some accelerated variants, have become popular for training in machine learning. However, in all existing SGD and its variants, the sample size in each iteration (epoch) of training is the same as the size of the full training set. In this paper, we propose a new method, called adaptive sample selection (ADASS), for training acceleration. During different epoches of training, ADASS only need to visit different training subsets which are adaptively selected from the full training set according to the Lipschitz constants of the loss functions on samples. It means that in ADASS the sample size in each epoch of training can be smaller than the size of the full training set, by discarding some samples. ADASS can be seamlessly integrated with existing optimization methods, such as SGD and momentum SGD, for training acceleration. Theoretical results show that the learning accuracy of ADASS is comparable to that of counterparts with full training set. Furthermore, empirical results on both shallow models and deep models also show that ADASS can accelerate the training process of existing methods without sacrificing accuracy.

READ FULL TEXT VIEW PDF
POST COMMENT

Comments

There are no comments yet.

Authors

page 1

page 2

page 3

page 4

07/28/2020

Stochastic Normalized Gradient Descent with Momentum for Large Batch Training

Stochastic gradient descent (SGD) and its variants have been the dominat...
02/26/2021

On the Generalization of Stochastic Gradient Descent with Momentum

While momentum-based methods, in conjunction with stochastic gradient de...
05/29/2019

Accelerated Sparsified SGD with Error Feedback

We study a stochastic gradient method for synchronous distributed optimi...
09/01/2021

The emergence of a concept in shallow neural networks

We consider restricted Boltzmann machine (RBMs) trained over an unstruct...
06/01/2011

Committee-Based Sample Selection for Probabilistic Classifiers

In many real-world learning tasks, it is expensive to acquire a sufficie...
05/22/2017

Large Scale Empirical Risk Minimization via Truncated Adaptive Newton Method

We consider large scale empirical risk minimization (ERM) problems, wher...
10/26/2018

Efficient Distributed Hessian Free Algorithm for Large-scale Empirical Risk Minimization via Accumulating Sample Strategy

In this paper, we propose a Distributed Accumulated Newton Conjugate gra...
This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.

1 Introduction

Many machine learning models can be formulated as the following empirical risk minimization problem:

(1)

where is the parameter to learn, corresponds to the loss on the th training sample, and is the total number of training samples.

With the rapid growth of data in real applications, stochastic optimization methods have become more popular than batch ones to solve the problem in (1). The most popular stochastic optimization method is stochastic gradient decent (SGD) (Zhang, 2004; Xiao, 2009; Bottou, 2010; Duchi et al., 2010). One practical way to adopt SGD for learning is the so-called Epoch-SGD  (Hazan and Kale, 2014) in Algorithm 1

, which has been widely used by mainstream machine learning platforms like Pytorch and TensorFlow. In each outer iteration (also called epoch) of Algorithm 

1, Epoch-SGD first samples a sequence from according to a distribution

defined on the full training set. A typical distribution is the uniform distribution. We can also set the sequence to be a permutation of

 (Tseng, 1998). In the inner iteration, the stochastic gradients computed based on the sampled sequence will be used to update the parameter. The mini-batch size is one in the inner iteration of Algorithm 1. In real applications, larger mini-batch size can also be used. After the inner iteration is completed, Epoch-SGD adjusts the step size to guarantee that is a non-increasing sequence. In general, we take . Although many theoretical results suggest to be the average of , we usually take the last one to be the initialization of the next outer iteration.

1:  Initialization: ;
2:  for  do
3:     Let be a sequence sampled from according to a distribution defined on the full training set;
4:     ;
5:     for  do
6:        ;
7:     end for
8:     ;
9:     Adjust step size to get ;
10:  end for
Algorithm 1 Epoch-SGD

To further accelerate Epoch-SGD in Algorithm 1, three main categories of methods have recently been proposed. The first category is to adopt momentum, Adam or Nesterov’s acceleration (Nesterov, 2007; Leen and Orr, 1993; Tseng, 1998; Lan, 2012; Kingma and Ba, 2014; Ghadimi and Lan, 2016; Allen-Zhu, 2018) to modify the update rule of SGD in Line 6 of Algorithm 1. This category of methods has faster convergence rate than SGD when

is small, and empirical results show that these methods are more stable than SGD. However, due to the variance of

, the convergence rate of these methods is the same as that in SGD when is large.

The second category is to design new stochastic gradients to replace in the inner iteration of Algorithm 1 such that the variance in the stochastic gradients can be reduced (Johnson and Zhang, 2013; Shalev-Shwartz and Zhang, 2013; Nitanda, 2014; Shalev-Shwartz and Zhang, 2014; Defazio et al., 2014; Schmidt et al., 2017). Representative methods include SAG (Schmidt et al., 2017) and SVRG (Johnson and Zhang, 2013). These methods can achieve faster convergence rate than vanilla SGD in most cases. However, the faster convergence of these methods are typically based on a smooth assumption for the objective function, which might not be satisfied in real problems. Another disadvantage of these methods is that they usually need extra memory cost and computation cost to get the stochastic gradients.

The third category is the importance sampling based methods, which try to design the distribution  (Zhao and Zhang, 2015; Csiba et al., 2015; Namkoong et al., 2017; Katharopoulos and Fleuret, 2018; Borsos et al., 2018). With properly designed distribution , these methods can also reduce the variance of and hence achieve faster convergence rate than SGD. (Zhao and Zhang, 2015) designs a distribution according to the global Lipschitz or smoothness. The distribution is firstly calculated based on the training set and then is fixed during the whole training process. (Csiba et al., 2015; Namkoong et al., 2017) proposes an adaptive distribution which will change in each epoch. (Borsos et al., 2018)

adopts online optimization to get the adaptive distribution. There also exist some other heuristic importance sampling methods 

(Shrivastava et al., 2016; Lin et al., 2017), which mainly focus on training samples with large loss (hard examples) and set the weight of samples with small loss to be small or .

One shortcoming of SGD and its variants, including the accelerated variants introduced above, is that the sample size in each iteration (epoch) of training is the same as the size of the full training set. This can also be observed in Algorithm 1, where a sequence of indices must be sampled from . Even for the importance sampling based methods, each sample in the full training set has possibility to be sampled in each outer iteration (epoch) and hence no samples can be discarded during training.

In this paper, we propose a new method, called adaptive sample selection (ADASS), to solve the above shortcoming of existing SGD and its variants. The contributions of ADASS are outlined as follows:

  • During different epoches of training, ADASS only need to visit different training subsets which are adaptively selected from the full training set according to the Lipschitz constants of the loss functions on samples. It means that in ADASS the sample size in each epoch of training can be smaller than the size of the full training set, by discarding some samples.

  • ADASS can be seamlessly integrated with existing optimization methods, such as SGD and momentum SGD, for training acceleration.

  • Theoretical results show that the learning accuracy of ADASS is comparable to that of counterparts with full training set.

  • Empirical results on both shallow models and deep models also show that ADASS can accelerate the training process of existing methods without sacrificing accuracy.

Notation: We use boldface lowercase letters like

to denote vectors, and use boldface uppercase letters like

to denote matrices. denotes the optimum of in (1) and denotes norm. , , with the th element being and others being . One epoch means that the algorithm passes through the selected training samples once.

2 A Simple Case: Least Square

We first adopt least square to give some hints for designing effective sample selection strategies, because least square is a simple model with closed-form solution.

Given a training set , where and . Least square tries to optimize the following objective function:

(2)

For convenience, let . Furthermore, we assume , which is generally satisfied when . Then, the optimal parameter of (2) can be directly computed as follows:

(3)

Let be a permutation of , and . Then, and denote the features and supervised information of the selected samples indexed by . For simplicity, we assume . Then it is easy to get that

(4)

We are interested in the difference between and

. If the difference is very small, it means that we can use less training samples to estimate

. We have the following lemma about the relationship between and . and satisfy the following equation:

Let . Based on Lemma 2, we can get the following corollary. Assume . Let be a permutation of , and . If such that

with the smallest eigenvalue

, when , we have

Here, actually corresponds to the Lipschitz constant of around . We call the bound of the first inequality in Corollary 2 loss bound because it is related to the loss on the samples. And we call the bound of the second inequality in Corollary 2 Lipschitz bound because it is related to the Lipschitz constants of the loss functions on the samples. We can find that in both loss bound and Lipschitz bound, the terms on the righthand side of the two inequalities correspond to those discarded (un-selected) training samples indexed by . Corollary 2 gives us a hint that in least square, to make the gap between and as small as possible, we should discard training samples with the smallest losses or smallest Lipschitz constants. That means we should select training samples with the largest losses or largest Lipschitz constants.

We design an experiment to further illustrate the results in Corollary 2. The feature is constructed from three different distributions: uniform distribution

, gaussian distribution

and binomial distribution

. The corresponding

is got by a linear transformation on

with a small gaussian noise. We compare three sample selection criterions: Lipschitz criterion according to Lipschitz bound, loss criterion according to loss bound, and random criterion with which samples are randomly selected. The result is shown in Figure 1, in which the y-axis denotes and the x-axis denotes the sampling ratio . We can find that both Lipschitz criterion and loss criterion achieve better performance than random criterion, for estimating with a subset of samples.

(a) uniform distribution
(b) gaussian distribution
(c) binomial distribution
Figure 1: Empirical illustration to compare different sample section criterions for least square.

3 Deep Analysis of Sample Selection Criterions

Based on the results of Corollary 2 about least square, it seems that both loss and Lipschitz constant can be adopted as criterions for sample selection. In this section, we give deep analysis about these two criterions and find that for general cases Lipschitz constant can be used for sample selection but loss cannot.

3.1 Loss based Sample Selection

Based on the loss criterion, in each iteration, the algorithm will select samples with the largest loss at current and learn with these selected samples to update . Intuitively, if the loss is large, it means the model has not fitted the th sample good enough and this sample need to be trained again.

Unfortunately, the loss based sample selection cannot theoretically guarantee the convergence of the learning procedure. We can give a negative example as follows: let . If we start from , then we will get . It means is a divergent sequence. In fact, even minimizes where is the selected samples with the largest loss at the th iteration, it can also make the other un-selected sample loss functions increase.

The loss criterion has another disadvantage. Let , and define

The can be treated as some unknown noise. It is easy to find that , minimizing is equivalent to minimizing . However, can disrupt the samples in seriously. In Figure 1, it is possible to design suitable to make the blue line be the same as the green line. Hence, the loss criterion is also not robust for sample selection.

3.2 Lipschtiz Constant based Sample Selection

In this subsection, we theoretically prove that Lipschtiz constant is a good criterion for sample selection.

Assume . Let . We say is -insignificant on if

(5)

where . This definition can be explained as follows. Inspired by the empirical result in Section 2 and the negative example in Section 3.1, we realize that the selected samples can only be used on a local region. So the definition above focuses on the local region . The right side of (5) denotes the decreased value of , and the left side is the increased value of . Mathematically, if is -insignificant on , we can obtain

(6)

Since , the above equation implies that with initialization , we can make decrease by minimizing on the local region. So when we optimize on a local region, plays a leading role, and has insignificant effect and subsequently can be discarded.

One trivial decomposition of is , where . Then for any , , is -insignificant on . In the following content, we will design a non-trivial decomposition of to facilitate sample selection. First, we give the following assumptions.

Assumption 1

(Local Lipschitz continuous) , there exists a constant such that ,

(7)
Assumption 2

For any fixed , there exists a constant such that , with ,

(8)

where is defined as .

For most loss functions used in machine learning, their gradients are bounded by a bounded closed domain which guarantees the Lipschitz continuous property. Hence, Assumption 1 is satisfied by most machine learning models. is the local Lipschitz constant which is determined by the specific sample and the neighborhood size of . Hence, we set different Lipschitz constants for different samples.

Assumption 2 can also be satisfied by most machine learning models, which is explained as follows. Let be the function defined in Assumption 2, and . If each is -strongly convex and , then such satisfy (8) with . Lemma 3.2 implies that Assumption 2 can be easily satisfied for strongly convex objective functions. For convex objective functions, it is easy to transform them to strongly convex objective functions by adding a small norm.

For non-convex objective functions, it is difficult to validate (8). However, we can empirically verify it. We randomly choose a and fix it. Then with the initialization , we train ResNet20 on a subset of cifar10 with the size  (). We run momentum SGD to estimate the local minimum around . For each , we repeat experiments 10 times. The result is in Figure 2. We can find that is almost proportional to . Hence, Assumption 2 is also reasonable for non-convex objective functions.

Figure 2: Empirical result for Assumption 2 on non-convex cases. The x-axis denotes , which is the size of subset . The y-axis represents the value .

With the above two assumptions, we can obtain the following theorem. Let with , and . We define two functions as follows:

Let

where satisfy Assumption 2. Then is -insignificant on . For any , according to Assumption 1 , we have

(9)

Since satisfy Assumption 2, according to the definition of , we have

Since , is -insignificant on .

According to the definition of in Theorem 3.2, to make , we firstly set with . If , then . Secondly, we can select samples with largest to construct and proper so that can be smaller than . It is reasonable because small local Lipschitz constant implies that the landscape of the loss function is flat and will not change too much on a local region.

The defined in Theorem 3.2 has another important property, which is stated in the following theorem. Assume are fixed, and assume and are defined as those in Theorem 3.2 with . Then minimizing is comparable to minimizing on with initialization , which means such that

where

(10)
(11)

Specifically, , .

Since is -insignificant on , we have

which means

On the other hand, according to Assumption 2, we have

Then we have

According to Theorem 3.2, we can directly get that

(12)

Moreover, we have the following corollary: With the defined in Theorem 3.2, we have

(13)

Moreover, if and let , we have

Here are defined in (10) and (11). It implies that for any , on the local region , we should select samples with the largest Lipschitz constants so that the gap between and can be guaranteed to be small. Since is obtained from the subset , we can use less training samples for optimization. This is also consistent with the second inequality in Corollary 2. Please note that Corollary 2 is only for least square, but the result in Corollary 3.2 can be used for general cases of machine learning.

4 Adass

From the empirical results in Figure 1, we can find that if we discard some training samples permanently which means that some samples are not used during the whole training process, it is difficult to get the optimal solution of (1). Hence, different training subsets need to be adaptively selected from the full training set for different iterations (epoches). Because sample selection of our method is adaptive to different training states, we name our method adaptive sample slection (ADASS). ADASS can be seamlessly integrated with any optimization methods, such as SGD or momentum SGD, to accelerate the training of these methods.

ADASS adopts Lipschitz constant as criterion for sample selection. To reduce the sample selection cost, we periodically perform sample selection after every epochs in our implementation. Furthermore, we use to estimate the local Lipschitz constant. Then, the criterion for sample selection is to choose a subset of the largest such that

Here, is a hyper-parameter denoting the threshold, denotes the selected samples in the th iteration, and has been omitted because both denominator and numerator have this common term.

The training procedure with ADASS is briefly listed in Algorithm 2. First, we train iterations with full training set to get . Then if the epoch satisfies , ADASS will select training samples adaptively. It means that after every epochs, ADASS will re-select training samples. Here, we adopt SGD and momentum SGD in Algorithm 2 for optimization. It is easy to see that when , ADASS is equal to vanila SGD or momentum SGD with full training set.

1:  Initialization: ;
2:  Train iterations with full training set, to get ;
3:  for  do
4:     if  then
5:        Select samples such that
(14)
6:     else
7:        ;
8:     end if
9:     Let be a sequence sampled from , and ,
10:     for  do
11:        Option I (SGD):
12:        Option II (momentum SGD):
13:     end for
14:     ;
15:     Adjust step size to get ;
16:  end for
Algorithm 2 Adaptive Sample Selection (ADASS)

4.1 Time Cost

In ADASS, it need to calculate the value

for sample selection. In large scale machine learning applications, this step will bring extra computation cost which can not be ignored. Here, we use deep neural network training on GPU as an example for demonstration. Let

  • : the time of loading one training sample;

  • : the time of calculating loss value of one training sample (forward);

  • : the time of calculating gradient of one training sample (backward).

When , ADASS is equal to SGD or momentum SGD so that we do not need to calculate loss value in Line 5 of Algorithm 2 for sample selection. The time cost of epochs of SGD is . When , we need to pass through the full training set once after every epochs for computing loss values. So the time cost of passing through with times of ADASS is . It is easy to get that

When , the time cost of ADASS is smaller than those of SGD and momentum SGD. For example, if , we can find that when , ADASS can accelerate the training process. However, we do not recommend to use large since the estimation error for Lipschitz constant might become larger with larger .

Figure 3:

Train logistic regression on mnist. The sampling ratio is defined as

. For FS, each epoch contains times of gradient computation. For ADASS and RS, the th epoch contains times of gradient computation.
Figure 4: Train ResNet20 on cifar10.

5 Experiments

We conduct experiments to evaluate ADASS with Option II in Algorithm 2. All the experiments are conducted on the Pytorch platform with GPU Titan XP.

We compare ADASS with two baselines:

  • FS: Full training samples without discarding samples. This corresponds to vanilla stochastic methods.

  • RS: Random sample selection. RS does not execute (14) in Algorithm 2. It randomly picks up indices from with the size being the same as of ADASS and during each epochs, the selected sample set is fixed.

For ADASS, we set . are trained (initialized) with FS. Hence, in all experiments, ADASS is the same as FS in the first 10 epochs. For all the three methods, we use momentum SGD as update rule.

5.1 Shallow Model (Convex)

The first experiment is to evaluate logistic regression (LR) with norm regularization, which is a shallow model and convex. The regularization coefficient is . The data set for evaluation is mnist 111http://yann.lecun.com/exdb/mnist/. We simply set for ADASS, and run three methods for 100 epochs ( in Algorithm 2).

The result is shown in Figure 3. The left figure shows the average training loss on the full training set (i.e., empirical risk). The middle one shows the test accuracy. We can find that ADASS (red line) and FS (blue line) achieve almost the same accuracy. However, RS achieves worse accuracy on both training loss and test accuracy. It implies that the training samples selected by ADASS are effective. The right figure shows sampling ratio in ADASS, which is defined as , during the whole training process. When , ADASS only need to select about half of training set. This means that ADASS can accelerate the training process of momentum SGD without sacrificing accuracy.

5.2 Deep Model (Non-Convex)

The second experiment is for deep models which are non-convex. Firstly, we evaluate ResNet20 on cifar10. The parameters for batch size, learning rate and momentum are the same as those in (He et al., 2016). We set for ADASS, and run the three methods for 160 epochs ( in Algorithm 2). The result is shown in Figure 4. We can find that ADASS (red line) and FS (blue line) achieve almost the same accuracy. RS achieves worse accuracy than FS, especially for training loss. More detailed results on ResNet20 are listed in Table 1. We can find that ADASS with achieves the best result on both training loss and test accuracy, by reducing about 30% time cost compared to FS (standard momentum SGD).

method c time (second) sampling ratio training loss test accuracy
FS 1 984.4 100% 0.016 91.4%
RS - 463.4 20.6% 0.09 90.5%
RS - 593.0 33.2% 0.059 90.8%
ADASS 0.99 511.0 20.6% 0.019 90.8%
ADASS 0.999 637.8 33.2% 0.016 91.4%
Table 1: Train ResNet20 on cifar10. The time in the table refers to the total time which includes sample selection cost. The sampling ratio is the ratio in the last epoch.

We also conduct experiments on the large dataset ImageNet using two models: ResNet18 and ResNet50. The parameters for batch size, learning rate and momentum are the same as those in

(He et al., 2016). The result is shown in Figure 5. We can find that with less training samples, ADASS gets almost the same accuracy as that of FS.

The experiments on deep models also show that ADASS can accelerate the training process of momentum SGD without sacrificing accuracy.

(a) Train ResNet18 on ImageNet
(b) Train ResNet50 on ImageNet
Figure 5: Train ResNet18 and ResNet50 on ImageNet.

6 Conclusion

In this paper, we propose a new method, called ADASS, for training acceleration. In ADASS, the sample size in each epoch of training can be smaller than the size of the full training set, by adaptively discarding some samples. ADASS can be seamlessly integrated with existing optimization methods, such as SGD and momentum SGD, for training acceleration. Empirical results show that ADASS can accelerate the training process of existing methods without sacrificing accuracy.

References

  • Allen-Zhu [2018] Zeyuan Allen-Zhu. Katyusha X: practical momentum method for stochastic sum-of-nonconvex optimization. In Proceedings of the 35th International Conference on Machine Learning, pages 179–185, 2018.
  • Borsos et al. [2018] Zalan Borsos, Andreas Krause, and Kfir Y. Levy. Online variance reduction for stochastic optimization. In Conference On Learning Theory, pages 324–357, 2018.
  • Bottou [2010] Léon Bottou.

    Large-scale machine learning with stochastic gradient descent.

    In Proceedings of the 19th International Conference on Computational Statistics, 2010.
  • Csiba et al. [2015] Dominik Csiba, Zheng Qu, and Peter Richtárik.

    Stochastic dual coordinate ascent with adaptive probabilities.

    In Proceedings of the 32nd International Conference on Machine Learning, pages 674–683, 2015.
  • Defazio et al. [2014] Aaron Defazio, Francis R. Bach, and Simon Lacoste-Julien. SAGA: a fast incremental gradient method with support for non-strongly convex composite objectives. In Advances in Neural Information Processing Systems, pages 1646–1654, 2014.
  • Duchi et al. [2010] John C. Duchi, Elad Hazan, and Yoram Singer. Adaptive subgradient methods for online learning and stochastic optimization. In COLT 2010 - The 23rd Conference on Learning Theory, Haifa, Israel, June 27-29, 2010, pages 257–269, 2010.
  • Ghadimi and Lan [2016] Saeed Ghadimi and Guanghui Lan. Accelerated gradient methods for nonconvex nonlinear and stochastic programming. Math. Program., 156(1-2):59–99, 2016.
  • Hazan and Kale [2014] Elad Hazan and Satyen Kale. Beyond the regret minimization barrier: optimal algorithms for stochastic strongly-convex optimization. Journal of Machine Learning Research, 15(1):2489–2512, 2014.
  • He et al. [2016] Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In

    2016 IEEE Conference on Computer Vision and Pattern Recognition

    , pages 770–778, 2016.
  • Johnson and Zhang [2013] Rie Johnson and Tong Zhang. Accelerating stochastic gradient descent using predictive variance reduction. In Advances in Neural Information Processing Systems, pages 315–323, 2013.
  • Katharopoulos and Fleuret [2018] Angelos Katharopoulos and Franccois Fleuret.

    Not all samples are created equal: Deep learning with importance sampling.

    In Proceedings of the 35th International Conference on Machine Learning, pages 2530–2539, 2018.
  • Kingma and Ba [2014] Diederik P. Kingma and Jimmy Ba. Adam: A method for stochastic optimization. CoRR, abs/1412.6980, 2014.
  • Lan [2012] Guanghui Lan. An optimal method for stochastic composite optimization. Math. Program., 133(1-2):365–397, 2012.
  • Leen and Orr [1993] Todd K. Leen and Genevieve B. Orr. Optimal stochastic search and adaptive momentum. In Advances in Neural Information Processing Systems, pages 477–484, 1993.
  • Lin et al. [2017] Tsung-Yi Lin, Priya Goyal, Ross B. Girshick, Kaiming He, and Piotr Dollár. Focal loss for dense object detection. In International Conference on Computer Vision, pages 2999–3007, 2017.
  • Namkoong et al. [2017] Hongseok Namkoong, Aman Sinha, Steve Yadlowsky, and John C. Duchi. Adaptive sampling probabilities for non-smooth optimization. In Proceedings of the 34th International Conference on Machine Learning, pages 2574–2583, 2017.
  • Nesterov [2007] Yu. Nesterov. Gradient methods for minimizing composite objective function, 2007.
  • Nitanda [2014] Atsushi Nitanda. Stochastic proximal gradient descent with acceleration techniques. In Advances in Neural Information Processing Systems, pages 1574–1582, 2014.
  • Schmidt et al. [2017] Mark W. Schmidt, Nicolas Le Roux, and Francis R. Bach. Minimizing finite sums with the stochastic average gradient. Math. Program., 162(1-2):83–112, 2017.
  • Shalev-Shwartz and Zhang [2013] Shai Shalev-Shwartz and Tong Zhang. Stochastic dual coordinate ascent methods for regularized loss. Journal of Machine Learning Research, 14(1):567–599, 2013.
  • Shalev-Shwartz and Zhang [2014] Shai Shalev-Shwartz and Tong Zhang. Accelerated proximal stochastic dual coordinate ascent for regularized loss minimization. In Proceedings of the 31th International Conference on Machine Learning, pages 64–72, 2014.
  • Shrivastava et al. [2016] Abhinav Shrivastava, Abhinav Gupta, and Ross B. Girshick. Training region-based object detectors with online hard example mining. In Conference on Computer Vision and Pattern Recognition, pages 761–769, 2016.
  • Tseng [1998] Paul Tseng. An incremental gradient(-projection) method with momentum term and adaptive stepsize rule. SIAM Journal on Optimization, 8(2):506–531, 1998.
  • Xiao [2009] Lin Xiao. Dual averaging method for regularized stochastic learning and online optimization. In Advances in Neural Information Processing Systems, pages 2116–2124, 2009.
  • Zhang [2004] Tong Zhang. Solving large scale linear prediction problems using stochastic gradient descent algorithms. In Machine Learning, Proceedings of the Twenty-first International Conference, 2004.
  • Zhao and Zhang [2015] Peilin Zhao and Tong Zhang. Stochastic optimization with importance sampling for regularized loss minimization. In Proceedings of the 32nd International Conference on Machine Learning, pages 1–9, 2015.