I am trying to implement a binary classifier using logistic regression for data drawn from 2 point sets (classes y (-1, 1)). As seen below, we can use the parameter a to prevent overfitting.
Now I am not sure, how to choose the "good" value for a.
Another thing I am not sure about is how to choose a "good" convergence criterion for this sort of problem.
Value of 'a'
Choosing "good" things is a sort of meta-regression: pick any value for a that seems reasonable. Run the regression. Try again with a values larger and smaller by a factor of 3. If either works better than the original, try another factor of 3 in that direction -- but round it from 9x to 10x for readability.
You get the idea ... play with it until you get in the right range. Unless you're really trying to optimize the result, you probably won't need to narrow it down much closer than that factor of 3.
Data Set Partition
ML folks have spent a lot of words analysing the best split. The optimal split depends very much on your data space. As a global heuristic, use half or a bit more for training; of the rest, no more than half should be used for testing, the rest for validation. For instance, 50:20:30 is a viable approximation for train:test:validate.
Again, you get to play with this somewhat ... except that any true test of the error rate would be entirely new data.
Convergence
This depends very much on the characteristics of your empirical error space near the best solution, as well as near local regions of low gradient.
The first consideration is to choose an error function that is likely to be convex and have no flattish regions. The second is to get some feeling for the magnitude of the gradient in the region of a desired solution (normalizing your data will help with this); use this to help choose the convergence radius; you might want to play with that 3x scaling here, too. The final one is to play with the learning rate, so that it's scaled to the normalized data.
Does any of this help?
Related
I understood the way to compute the forward part in Deep learning. Now, I want to understand the backward part. Let's take X(2,2) as an example. The backward at the position X(2,2) can compute as the figure bellow
My question is that where is dE/dY (such as dE/dY(1,1),dE/dY(1,2)...) in the formula? How to compute it at the first iteration?
SHORT ANSWER
Those terms are in the final expansion at the bottom of the slide; they contribute to the summation for dE/dX(2,2). In your first back-propagation, you start at the end and work backwards (hence the name) -- and the Y values are the ground-truth labels. So much for computing them. :-)
LONG ANSWER
I'll keep this in more abstract, natural-language terms. I'm hopeful that the alternate explanation will help you see the large picture as well as sorting out the math.
You start the training with assigned weights that may or may not be at all related to the ground truth (labels). You move blindly forward, making predictions at each layer based on naive faith in those weights. The Y(i,j) values are the resulting meta-pixels from that faith.
Then you hit the labels at the end. You work backward, adjusting each weight. Note that, at the last layer, the Y values are the ground-truth labels.
At each layer, you mathematically deal with two factors:
How far off was this prediction?
How heavily did this parameter contribute to that prediction?
You adjust the X-to-Y weight by "off * weight * learning_rate".
When you complete that for layer N, you back up to layer N-1 and repeat.
PROGRESSION
Whether you initialize your weights with fixed or random values (I generally recommend the latter), you'll notice that there's really not much progress in the early iterations. Since this is slow adjustment from guess-work weights, it takes several iterations to get a glimmer of useful learning into the last layers. The first layers are still cluelessly thrashing at this point. The loss function will bounce around close to its initial values for a while. For instance, with GoogLeNet's image recognition, this flailing lasts for about 30 epochs.
Then, finally, you get some valid learning in the latter layers, the patterns stabilize enough that some consistency percolates back to the early layers. At this point, you'll see the loss function drop to a "directed experimentation" level. From there, the progression depends a lot on the paradigm and texture of the problem: some have a sharp drop, then a gradual convergence; others have a more gradual drop, almost an exponential decay to convergence; more complex topologies have additional sharp drops as middle or early phases "get their footing".
I have been using the training method proposed in the cifar10_multi_gpu_train example for (local) multi-gpu training, i.e., creating several towers and then average the gradient. However, I was wondering the following: What does happen if I just take the losses coming from the different GPUs, sum them up and then just apply gradient descent to that new loss.
Would that work? Probably this is a silly question, and there must be a limitation somewhere. So I would be happy if you could comment on this.
Thanks and best regards,
G.
It would not work with the sum. You would get a bigger loss and consequentially bigger and probably erroneous gradients. While averaging the gradients you get an average of the direction that the weights have to take in order to minimize the loss, but each single direction is the one computed for the exact loss value.
One thing that you can try is to run the towers independently and then average the weights from time to time, slower convergence rate but faster processing on each node.
I am trying to tune the hyper parameter i.e batch size in CNN.I have a computer of corei7,RAM 12GB and i am training a CNN network with CIFAR-10 dataset which can be found in this blog.Now At first what i have read and learnt about batch size in machine learning:
let's first suppose that we're doing online learning, i.e. that we're
using a minibatch size of 1. The obvious worry about online learning
is that using minibatches which contain just a single training
example will cause significant errors in our estimate of the gradient.
In fact, though, the errors turn out to not be such a problem. The
reason is that the individual gradient estimates don't need to be
superaccurate. All we need is an estimate accurate enough that our
cost function tends to keep decreasing. It's as though you are trying
to get to the North Magnetic Pole, but have a wonky compass that's
10-20 degrees off each time you look at it. Provided you stop to
check the compass frequently, and the compass gets the direction right
on average, you'll end up at the North Magnetic Pole just
fine.
Based on this argument, it sounds as though we should use online
learning. In fact, the situation turns out to be more complicated than
that.As we know we can use matrix techniques to compute the gradient
update for all examples in a minibatch simultaneously, rather than
looping over them. Depending on the details of our hardware and linear
algebra library this can make it quite a bit faster to compute the
gradient estimate for a minibatch of (for example) size 100 , rather
than computing the minibatch gradient estimate by looping over the
100 training examples separately. It might take (say) only 50 times as
long, rather than 100 times as long.Now, at first it seems as though
this doesn't help us that much.
With our minibatch of size 100 the learning rule for the weights
looks like:
where the sum is over training examples in the minibatch. This is
versus for online learning.
Even if it only takes 50 times as long to do the minibatch update, it
still seems likely to be better to do online learning, because we'd be
updating so much more frequently. Suppose, however, that in the
minibatch case we increase the learning rate by a factor 100, so the
update rule becomes
That's a lot like doing separate instances of online learning with a
learning rate of η. But it only takes 50 times as long as doing a
single instance of online learning. Still, it seems distinctly
possible that using the larger minibatch would speed things up.
Now i tried with MNIST digit dataset and ran a sample program and set the batch size 1 at first.I noted down the training time needed for the full dataset.Then i increased the batch size and i noticed that it became faster.
But in case of training with this code and github link changing the batch size doesn't decrease the training time.It remained same if i use 30 or 128 or 64.They are saying that they got 92% accuracy.After two or three epoch they have got above 40% accuracy.But when i ran the code in my computer without changing anything other than the batch size i got worse result after 10 epoch like only 28% and test accuracy stuck there in the next epochs.Then i thought since they have used batch size of 128 i need to use that.Then i used the same but it became more worse only give 11% after 10 epoch and stuck in there.Why is that??
Neural networks learn by gradient descent an error function in the weight space which is parametrized by the training examples. This means the variables are the weights of the neural network. The function is "generic" and becomes specific when you use training examples. The "correct" way would be to use all training examples to make the specific function. This is called "batch gradient descent" and is usually not done for two reasons:
It might not fit in your RAM (usually GPU, as for neural networks you get a huge boost when you use the GPU).
It is actually not necessary to use all examples.
In machine learning problems, you usually have several thousands of training examples. But the error surface might look similar when you only look at a few (e.g. 64, 128 or 256) examples.
Think of it as a photo: To get an idea of what the photo is about, you usually don't need a 2500x1800px resolution. A 256x256px image will give you a good idea what the photo is about. However, you miss details.
So imagine gradient descent to be a walk on the error surface: You start on one point and you want to find the lowest point. To do so, you walk down. Then you check your height again, check in which direction it goes down and make a "step" (of which the size is determined by the learning rate and a couple of other factors) in that direction. When you have mini-batch training instead of batch-training, you walk down on a different error surface. In the low-resolution error surface. It might actually go up in the "real" error surface. But overall, you will go in the right direction. And you can make single steps much faster!
Now, what happens when you make the resolution lower (the batch size smaller)?
Right, your image of what the error surface looks like gets less accurate. How much this affects you depends on factors like:
Your hardware/implementation
Dataset: How complex is the error surface and how good it is approximated by only a small portion?
Learning: How exactly are you learning (momentum? newbob? rprop?)
I'd like to add to what's been already said here that larger batch size is not always good for generalization. I've seen these cases myself, when an increase in batch size hurt validation accuracy, particularly for CNN working with CIFAR-10 dataset.
From "On Large-Batch Training for Deep Learning: Generalization Gap and Sharp Minima":
The stochastic gradient descent (SGD) method and its variants are
algorithms of choice for many Deep Learning tasks. These methods
operate in a small-batch regime wherein a fraction of the training
data, say 32–512 data points, is sampled to compute an approximation
to the gradient. It has been observed in practice that when using a
larger batch there is a degradation in the quality of the model, as
measured by its ability to generalize. We investigate the cause for
this generalization drop in the large-batch regime and present
numerical evidence that supports the view that large-batch methods
tend to converge to sharp minimizers of the training and testing
functions—and as is well known, sharp minima lead to poorer
generalization. In contrast, small-batch methods consistently converge
to flat minimizers, and our experiments support a commonly held view
that this is due to the inherent noise in the gradient estimation. We
discuss several strategies to attempt to help large-batch methods
eliminate this generalization gap.
Bottom-line: you should tune the batch size, just like any other hyperparameter, to find an optimal value.
The 2018 opinion retweeted by Yann LeCun is the paper Revisiting Small Batch Training For Deep Neural Networks, Dominic Masters and Carlo Luschi suggesting a good generic maximum batch size is:
32
With some interplay with choice of learning rate.
The earlier 2016 paper On Large-batch Training For Deep Learning: Generalization Gap And Sharp Minima gives some reason for not using big batches, which I paraphrase badly, as big batches are likely to get stuck in local (“sharp”) minima, small batches not.
I have some basic conceptual queries on SVM - it will be great if any one can guide me on this. I have been studying books and lectures for a while but have not been able to get answers for these queries correctly
Suppose I have m featured data points - m > 2. How will I know if the data points are linearly separable or not?. If I have understood correctly, linearly separable data points - will not need any special kernel for finding the hyper plane as there is no need to increase the dimension.
Say, I am not sure whether the data is linearly separable or not. I try to get a hyper plane with linear kernel, once with slackness and once without slackness on the lagrange multipliers. What difference will I see on the error rates on training and test data for these two hyper planes. If I understood correctly, if the data is not linearly separable, and if I am not using slackness then there cannot be any optimal plane. If that is the case, should the svm algorithm give me different hyper planes on different runs. Now when I introduce slackness - should I always get the same hyper plane, every run ? And how exactly can I find out from the lagrange multipliers of a hyper plane, whether the data was linearly separable or not.
Now say from 2 I came to know somehow that the data was not linearly separable at m dimensions. So I will try to increase the dimensions and see if it is separable at a higher dimension. How do I know how high I will need to go ? I know the calculations do not go into that space - but is there any way to find out from 2 what should be the best kernel for 3 (i.e I want to find a linearly separating hyper plane).
What is the best way to visualize hyper planes and data points in Matlab where the feature dimensions can be as big as 60 - and the hyperplane is at > 100 dimensions (i,e data points in few hundreds and using Gaussian Kernels the feature vector changes to > 100 dimensions).
I will really appreciate if someone clears these doubts
Regards
I'm going to try to focus on your questions (1), (2) and (3). In practice the most important concern is not if the problem becomes linearly separable but how well the classifier performs on unseen data (i.e. how well it classifies). It seems you want to find a good kernel for which data is linearly separable, and you will always be able to do this (consider putting at each training point an extremely narrow gaussian RBF), but what you really want is good performance on unseen data. That being said:
If the problem is not linearly separable and not using slacks the optimization will fail. It depends on the implementation and the specific optimization algorithm how it fails, does it not converge?, does it not find a descent direction? does it run into numerical difficulties? Even if you wanted to determine cases with slacks, you can still run into numerical difficulties that would make and that alone would make your algorithm of linear separability unreliable
How high do you need to go? Well that is a fundamental question. It is called the problem of data representation. For straight forward solutions people use held out data (people don't care about linear separability they care about good performance on held out data) and do parameter search (for example an RBF kernel can is strictly more expressive than a linear kernel) under the correct gammas. So the problem becomes finding a good gamma for your data. See for example this paper: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.141.880
I don't think there is a trivial connection between the values of the lagrangian multipliers and linear separability. You can try an high alphas whose value is C, but I'm not sure you'll be able to say much.
Trying to write some code that deals with this task:
As an starting point, I have around 20 "profiles" (imagine a landscape profile), i.e. one-dimensional arrays of around 1000 real values.
Each profile has a real-valued desired outcome, the "effective height".
The effective height is some sort of average but height, width and position of peaks play a particular role.
My aim is to generalize from the input data so as to calculate the effective height for further profiles.
Is there a machine learning algorithm or principle that could help?
Principle 1: Extract the most import features, instead of feeding it everything
As you said, "The effective height is some sort of average but height, width and position of peaks play a particular role." So that you have a strong priori assumption that these measures are the most important for learning. If I were you, I would calculate these measures at first, and use them as the input for learning, instead of the raw data.
Principle 2: While choosing a learning algorithm, the first thing to care about would be the the linear separability
Suppose the height is a function of those measures, then you have to think about that to what extent the function is linear. For example if the function is almost linear, then a very simple Perceptron would be perfect. Otherwise if it's far from linear, you might want to pick up a multiple-layer neural network. If it's far far far from linear....please turn to principle 1, and check out if you are extracting the right features.
Principle 3: More data help
As you said, you have around 20 "profiles" for training. In general speaking, that's not enough. Almost all of the machine learning algorithms were designed for somehow big data. Even they claimed that their algorithm is good at learning small sample, but usually not as small as 20. Get more data!
Maybe multivariate linear regression suffices?
I would probably use a combination of what you said about which features play the most important role, and then train a regression on that. Basically, you need at least one coefficient corresponding to each feature, and you need substantially more data points than coefficients. So, I would pick something like the heights and width of the two biggest peaks. You've now reduced every profile to just 4 numbers. Now do this trick: divide the data into 5 groups of 4. Pick the first 4 groups. Reduce all those profiles to 4 numbers, and then use the desired outcomes to come up with a regression. Once you have trained the regression, try your technique on the last 4 points and see how well it works. Repeat this procedure 5 times, each time leaving out a different set of data. This is called cross-validation, and it's very handy.
Obviously getting more data would help.