This post explores how many of the most popular gradient-based optimization algorithms actually work.
Note: If you are looking for a review paper, this blog post is also available as an article on arXiv.
Update 20.03.2020: Added a note on recent optimizers.
Update 09.02.2018: Added AMSGrad.
Update 24.11.2017: Most of the content in this article is now also available as slides.
Update 15.06.2017: Added derivations of AdaMax and Nadam.
Update 21.06.16: This post was posted to Hacker News. The discussion provides some interesting pointers to related work and other techniques.
Table of contents:
- Gradient descent variants
- Challenges
- Gradient descent optimization algorithms
- Parallelizing and distributing SGD
- Additional strategies for optimizing SGD
- Conclusion
- References
Gradient descent is one of the most popular algorithms to perform optimization and by far the most common way to optimize neural networks. At the same time, every state-of-the-art Deep Learning library contains implementations of various algorithms to optimize gradient descent (e.g. lasagne’s, caffe’s, and keras’ documentation). These algorithms, however, are often used as black-box optimizers, as practical explanations of their strengths and weaknesses are hard to come by.
This blog post aims at providing you with intuitions towards the behaviour of different algorithms for optimizing gradient descent that will help you put them to use. We are first going to look at the different variants of gradient descent. We will then briefly summarize challenges during training. Subsequently, we will introduce the most common optimization algorithms by showing their motivation to resolve these challenges and how this leads to the derivation of their update rules. We will also take a short look at algorithms and architectures to optimize gradient descent in a parallel and distributed setting. Finally, we will consider additional strategies that are helpful for optimizing gradient descent.
Gradient descent is a way to minimize an objective function (J(theta)) parameterized by a model’s parameters (theta in mathbb{R}^d ) by updating the parameters in the opposite direction of the gradient of the objective function (nabla_theta J(theta)) w.r.t. to the parameters. The learning rate (eta) determines the size of the steps we take to reach a (local) minimum. In other words, we follow the direction of the slope of the surface created by the objective function downhill until we reach a valley. If you are unfamiliar with gradient descent, you can find a good introduction on optimizing neural networks here.
There are three variants of gradient descent, which differ in how much data we use to compute the gradient of the objective function. Depending on the amount of data, we make a trade-off between the accuracy of the parameter update and the time it takes to perform an update.
Batch gradient descent
Vanilla gradient descent, aka batch gradient descent, computes the gradient of the cost function w.r.t. to the parameters (theta) for the entire training dataset:
(theta = theta – eta cdot nabla_theta J( theta)).
As we need to calculate the gradients for the whole dataset to perform just one update, batch gradient descent can be very slow and is intractable for datasets that don’t fit in memory. Batch gradient descent also doesn’t allow us to update our model online, i.e. with new examples on-the-fly.
In code, batch gradient descent looks something like this:
for i in range(nb_epochs):
params_grad = evaluate_gradient(loss_function, data, params)
params = params - learning_rate * params_grad
For a pre-defined number of epochs, we first compute the gradient vector params_grad of the loss function for the whole dataset w.r.t. our parameter vector params. Note that state-of-the-art deep learning libraries provide automatic differentiation that efficiently computes the gradient w.r.t. some parameters. If you derive the gradients yourself, then gradient checking is a good idea. (See here for some great tips on how to check gradients properly.)
We then update our parameters in the opposite direction of the gradients with the learning rate determining how big of an update we perform. Batch gradient descent is guaranteed to converge to the global minimum for convex error surfaces and to a local minimum for non-convex surfaces.
Stochastic gradient descent
Stochastic gradient descent (SGD) in contrast performs a parameter update for each training example (x^{(i)}) and label (y^{(i)}):
(theta = theta – eta cdot nabla_theta J( theta; x^{(i)}; y^{(i)})).
Batch gradient descent performs redundant computations for large datasets, as it recomputes gradients for similar examples before each parameter update. SGD does away with this redundancy by performing one update at a time. It is therefore usually much faster and can also be used to learn online.
SGD performs frequent updates with a high variance that cause the objective function to fluctuate heavily as in Image 1.
While batch gradient descent converges to the minimum of the basin the parameters are placed in, SGD’s fluctuation, on the one hand, enables it to jump to new and potentially better local minima. On the other hand, this ultimately complicates convergence to the exact minimum, as SGD will keep overshooting. However, it has been shown that when we slowly decrease the learning rate, SGD shows the same convergence behaviour as batch gradient descent, almost certainly converging to a local or the global minimum for non-convex and convex optimization respectively.
Its code fragment simply adds a loop over the training examples and evaluates the gradient w.r.t. each example. Note that we shuffle the training data at every epoch as explained in this section.
for i in range(nb_epochs):
np.random.shuffle(data)
for example in data:
params_grad = evaluate_gradient(loss_function, example, params)
params = params - learning_rate * params_gradMini-batch gradient descent
Mini-batch gradient descent finally takes the best of both worlds and performs an update for every mini-batch of (n) training examples:
(theta = theta – eta cdot nabla_theta J( theta; x^{(i:i+n)}; y^{(i:i+n)})).
This way, it a) reduces the variance of the parameter updates, which can lead to more stable convergence; and b) can make use of highly optimized matrix optimizations common to state-of-the-art deep learning libraries that make computing the gradient w.r.t. a mini-batch very efficient. Common mini-batch sizes range between 50 and 256, but can vary for different applications. Mini-batch gradient descent is typically the algorithm of choice when training a neural network and the term SGD usually is employed also when mini-batches are used. Note: In modifications of SGD in the rest of this post, we leave out the parameters (x^{(i:i+n)}; y^{(i:i+n)}) for simplicity.
In code, instead of iterating over examples, we now iterate over mini-batches of size 50:
for i in range(nb_epochs):
np.random.shuffle(data)
for batch in get_batches(data, batch_size=50):
params_grad = evaluate_gradient(loss_function, batch, params)
params = params - learning_rate * params_gradVanilla mini-batch gradient descent, however, does not guarantee good convergence, but offers a few challenges that need to be addressed:
-
Choosing a proper learning rate can be difficult. A learning rate that is too small leads to painfully slow convergence, while a learning rate that is too large can hinder convergence and cause the loss function to fluctuate around the minimum or even to diverge.
-
Learning rate schedules try to adjust the learning rate during training by e.g. annealing, i.e. reducing the learning rate according to a pre-defined schedule or when the change in objective between epochs falls below a threshold. These schedules and thresholds, however, have to be defined in advance and are thus unable to adapt to a dataset’s characteristics .
-
Additionally, the same learning rate applies to all parameter updates. If our data is sparse and our features have very different frequencies, we might not want to update all of them to the same extent, but perform a larger update for rarely occurring features.
-
Another key challenge of minimizing highly non-convex error functions common for neural networks is avoiding getting trapped in their numerous suboptimal local minima. Dauphin et al. argue that the difficulty arises in fact not from local minima but from saddle points, i.e. points where one dimension slopes up and another slopes down. These saddle points are usually surrounded by a plateau of the same error, which makes it notoriously hard for SGD to escape, as the gradient is close to zero in all dimensions.
In the following, we will outline some algorithms that are widely used by the deep learning community to deal with the aforementioned challenges. We will not discuss algorithms that are infeasible to compute in practice for high-dimensional data sets, e.g. second-order methods such as Newton’s method.
Momentum
SGD has trouble navigating ravines, i.e. areas where the surface curves much more steeply in one dimension than in another , which are common around local optima. In these scenarios, SGD oscillates across the slopes of the ravine while only making hesitant progress along the bottom towards the local optimum as in Image 2.
Momentum is a method that helps accelerate SGD in the relevant direction and dampens oscillations as can be seen in Image 3. It does this by adding a fraction (gamma) of the update vector of the past time step to the current update vector:
(
begin{align}
begin{split}
v_t &= gamma v_{t-1} + eta nabla_theta J( theta) \
theta &= theta – v_t
end{split}
end{align}
)
Note: Some implementations exchange the signs in the equations. The momentum term (gamma) is usually set to 0.9 or a similar value.
Essentially, when using momentum, we push a ball down a hill. The ball accumulates momentum as it rolls downhill, becoming faster and faster on the way (until it reaches its terminal velocity if there is air resistance, i.e. (gamma < 1)). The same thing happens to our parameter updates: The momentum term increases for dimensions whose gradients point in the same directions and reduces updates for dimensions whose gradients change directions. As a result, we gain faster convergence and reduced oscillation.
However, a ball that rolls down a hill, blindly following the slope, is highly unsatisfactory. We’d like to have a smarter ball, a ball that has a notion of where it is going so that it knows to slow down before the hill slopes up again.
Nesterov accelerated gradient (NAG) is a way to give our momentum term this kind of prescience. We know that we will use our momentum term (gamma v_{t-1}) to move the parameters (theta). Computing ( theta – gamma v_{t-1} ) thus gives us an approximation of the next position of the parameters (the gradient is missing for the full update), a rough idea where our parameters are going to be. We can now effectively look ahead by calculating the gradient not w.r.t. to our current parameters (theta) but w.r.t. the approximate future position of our parameters:
(
begin{align}
begin{split}
v_t &= gamma v_{t-1} + eta nabla_theta J( theta – gamma v_{t-1} ) \
theta &= theta – v_t
end{split}
end{align}
)
Again, we set the momentum term (gamma) to a value of around 0.9. While Momentum first computes the current gradient (small blue vector in Image 4) and then takes a big jump in the direction of the updated accumulated gradient (big blue vector), NAG first makes a big jump in the direction of the previous accumulated gradient
