I have trained a Convolution Neural Network, after comparing two normalizations,
I found that simple minus mean and divided by standard variance is better than scaling into [0, 1], it seems that the interval of input value is unnecessary in [0, 1] with sigmoid function.
Does anybody could explain about it?
If you're talking about a NN using logistic regression, then you are correct that a suitable sigmoid function (or logistic function in this context) will give you a [0, 1] range from your original inputs.
However, the logistic function works best when the inputs are in a small range on either side of zero - so, for example, your input to the logistic function might be [-3, +3].
By rescaling your data to [0, 1] first, you would flatten out any underlying distribution and move all of your data to the positive side of zero, which is not what the logistic function expects. So you will get a worse result than by normalising (i.e. subtract mean and divide by standard deviation, as you said) because that normalisation step takes account of the variance in the original distribution and makes sure that the mean is zero so you get both positive and negative data input to the logistic function.
In your question, you said "comparing two normalisations" - I think you are misunderstanding what "normalisation" means and actually comparing normalisation with rescaling, which is different.
Related
I'm relatively new to Neural Networks.
Atm I am trying to program a Neural Network for simple image recognition of numbers between 0 and 10.
The activation function I'm aiming for is ReLU (rectified linear unit).
With the sigmoid-function it is pretty clear how you can determine a probability for a certain case in the end (because its between 0 and 1).
But as far as I understand it, with the ReLU we don't have these limitations, but can get any value as a sum of previous "neurons" in the end.
So how is this commonly solved?
Do I just take the biggest of all values and say thats probability 100%?
Do I sum up all values and say thats the 100%?
Or is there another aproach I can't see atm?
I hope my question is understandable.
Thanks in advance for taking the time, looking at my question.
You can't use ReLU function as the output function for classification tasks because, as you mentioned, its range can't represent probability 0 to 1. That's why it is used only for regression tasks and hidden layers.
For binary classification, you have to use output function with range between 0 to 1 such as sigmoid. In your case, you would need a multidimensional extension such as softmax function.
I'm a tad confused here. I just started on the subject of Neural Networks and the first one I constructed used the Step-Activation with thresholds on each neuron. Now I wan't to implement the sigmoid activation but it seems that this type of activation doesn't use thresholds, only weights between the neurons. But in the information I find about this there is word of thresholds, only I can't find where they should be in the activation function.
Are thresholds used in a sigmoid activation function in neural networks?
There is no discrete jump as in step activation. The threshold could be considered to be the point where the sigmoid function is 0.5. Some sigmoid functions will have this at 0, while some will have it set to a different 'threshold'.
The step function may be thought of as a version of the sigmoid function that has the steepness set to infinity. There is an obvious threshold in this case, and for less steep sigmoid functions, the threshold could be considered to be where the function's value is 0.5, or the point of maximum steepness.
Sigmoid function's value is in the range [0;1], 0.5 is taken as a threshold, if h(theta) < 0.5 we assume that it's value is 0, if h(theta) >= 0.5 then it's 1.
Thresholds are used only on the output layer of the network and it's only when classifying. So, if you're trying to classify between 4 classes, then the output layer has 4 nodes y = [y1,y2,y3,y4], you'll use this threshold to assign y[i] 1 or 0.
It doesn't need to. Sigmoid curve itself partially can act as a threshold.
I want to know whether Artificial Neural Networks can be applied to discrete values inputs? I know they can be applied to continuous valued inputs, but can they be applied to discrete valued ones? Also, will perform well for discrete valued inputs?
Yes, artificial neural networks may be applied to data featuring discrete-value input variables. In the most commonly used neural network architectures (which are numeric), discrete inputs are typically represented by a series of dummy variables, just as in statistical regression. Also, as with regression, one less than the number of distinct values dummy variables is needed. There are other methods, but this is the most straightforward.
Well, good question let me say!
First of all let me answer directly yes to your question!
The answer implies to consider few aspects about the use and the implementation of the network itself.
Than let me explain why:
The easiest way is to normalize input as usual, this is the first rule of thumb with NN,
than let the neural network compute the task, and once you have your output, invert the normalization to get the output in the original range but still continuous, to get back descrete values just consider the integer part of your output. It is easy, it works and is fine, DONE! A good result just depends on the topology you design for you network.
As a plus you could consider the use of "step" transfer function, instead of "tan-sigmoid", between layers just to strenght and mimic a sort of digitization forcing the output to be just 0 or 1. But you should reconsider also the starting normalization as well as the use of well tuned thresholds.
NB: this latter trick is not really necessary but could give some secondary benefits; maybe test it in a second stage of your development and look at the differences.
PS: Just let me suggest something that should apply to your issue; if you would be smart take into account the use of some fuzzy logic on your learning algorithm ;-)
Cheers!
I'm late on this question, but this may help someone.
Say you have a categorical output variable, for example 3 different categories (0, 1 and 2),
outputs
0
2
1
2
1
0
then becomes
1, 0, 0
0, 0, 1
0, 1, 0
0, 0, 1
0, 1, 0
1, 0, 0
A possible NN output result is
0.2, 0.3, 0.5 (winner is categ 2)
0.05, 0.9, 0.05 (winner is categ 1)
...
Then your NN hill have 3 output nodes in this case, so take the max value.
To improve this, use entropy as a error measure and a softmax activation on the output layer, so that the outputs sum up to 1.
The purpose of a neural network is to approximate complicated functions by interpolating samples. As such, they tend to be a poor fit for discrete data, unless that data can be expressed by thresholding a continuous function. Depending on your problem, there are likely to be much more effective learning methods.
Let's say I want to determine the probability that I will upvote a question on SO, based only on which tags are present or absent.
Let's also imagine that I have plenty of data about past questions that I did or did not upvote.
Is there a machine learning algorithm that could take this historical data, train on it, and then be able to predict my upvote probability for future questions? Note that it must be the probability, not just some arbitrary score.
Let's assume that there will be up-to 7 tags associated with any given question, these being drawn from a superset of tens of thousands.
My hope is that it is able to make quite sophisticated connections between tags, rather than each tag simply contributing to the end result in a "linear" way (much as words do in a Bayesian spam filter).
So for example, it might be that the word "java" increases my upvote probability, except when it is present with "database", however "database" might increase my upvote probability when present with "ruby".
Oh, and it should be computationally reasonable (training within an hour or two on millions of questions).
What approaches should I research here?
Given that there probably aren't many tags per message, you could just create "n-gram" tags and apply naive Bayes. Regression trees would also produce an empirical probability at the leaf nodes, using +1 for upvote and 0 for no upvote. See http://www.stat.cmu.edu/~cshalizi/350-2006/lecture-10.pdf for some readable lecture notes and http://sites.google.com/site/rtranking/ for an open source implementation.
You can try several methods (linear regression, SMV, neural networks). The input vector should consist of all possible tags, where each tag represents one dimension.
Then each record in a training set has to be transformed to the input vector according to the tags. For example let's say you have different combinations of 4 tags in your training set (php, ruby, ms, sql) and you define an unweighted input vector [php, ruby, ms, sql]. Let's say you have the following 3 records whic are transformed to weighted input vectors:
php, sql -> [1, 0, 0, 1]
ruby -> [0, 1, 0, 0]
ms, sql -> [0, 0, 1, 1]
In case you use linear regression you use the following formula
y = k * X
where y represents an answer (upvote/downvote) in your case and by inserting known values (X - weighted input vectors).
How ta calculate weights in case you use linear regression you can read here but the point is to create binary input vectors which size is equal (or larger in case you take into account some other variables) to the number of all tags and then for each record you set weights for each tag (0 if it is not included or 1 otherwise).
I've read about neural network a little while ago and I understand how an ANN (especially a multilayer perceptron that learns via backpropagation) can learn to classify an event as true or false.
I think there are two ways :
1) You get one output neuron. It it's value is > 0.5 the events is likely true, if it's value is <=0.5 the event is likely to be false.
2) You get two output neurons, if the value of the first is > than the value of the second the event is likely true and vice versa.
In these case, the ANN tells you if an event is likely true or likely false. It does not tell how likely it is.
Is there a way to convert this value to some odds or to directly get odds out of the ANN. I'd like to get an output like "The event has a 84% probability to be true"
Once a NN has been trained, for eg. using backprogation as mentioned in the question (whereby the backprogation logic has "nudged" the weights in ways that minimize the error function) the weights associated with all individual inputs ("outside" inputs or intra-NN inputs) are fixed. The NN can then be used for classifying purposes.
Whereby the math (and the "options") during the learning phase can get a bit thick, it is relatively simple and straightfoward when operating as a classifier. The main algorithm is to compute an activation value for each neuron, as the sum of the input x weight for that neuron. This value is then fed to an activation function which purpose's is to normalize it and convert it to a boolean (in typical cases, as some networks do not have an all-or-nothing rule for some of their layers). The activation function can be more complex than you indicated, in particular it needn't be linear, but whatever its shape, typically sigmoid, it operate in the same fashion: figuring out where the activation fits on the curve, and if applicable, above or below a threshold. The basic algorithm then processes all neurons at a given layer before proceeding to the next.
With this in mind, the question of using the perceptron's ability to qualify its guess (or indeed guesses - plural) with a percentage value, finds an easy answer: you bet it can, its output(s) is real-valued (if anything in need of normalizing) before we convert it to a discrete value (a boolean or a category ID in the case of several categories), using the activation functions and the threshold/comparison methods described in the question.
So... How and Where do I get "my percentages"?... All depends on the NN implementation, and more importantly, the implementation dictates the type of normalization functions that can be used to bring activation values in the 0-1 range and in a fashion that the sum of all percentages "add up" to 1. In its simplest form, the activation function can be used to normalize the value and the weights of the input to the output layer can be used as factors to ensure the "add up" to 1 question (provided that these weights are indeed so normalized themselves).
Et voilà!
Claritication: (following Mathieu's note)
One doesn't need to change anything in the way the Neural Network itself works; the only thing needed is to somehow "hook into" the logic of output neurons to access the [real-valued] activation value they computed, or, possibly better, to access the real-valued output of the activation function, prior its boolean conversion (which is typically based on a threshold value or on some stochastic function).
In other words, the NN works as previously, neither its training nor recognition logic are altered, the inputs to the NN stay the same, as do the connections between various layers etc. We only get a copy of the real-valued activation of the neurons in the output layer, and we use this to compute a percentage. The actual formula for the percentage calculation depends on the nature of the activation value and its associated function (its scale, its range relative to other neurons' output etc.).
Here are a few simple cases (taken from the question's suggested output rules)
1) If there is a single output neuron: the ratio of the value provided by the activation function relative to the range of that function should do.
2) If there are two (or more output neurons), as with classifiers for example: If all output neurons have the same activation function, the percentage for a given neuron is that of its activation function value divided by the sum of all activation function values. If the activation functions vary, it becomes a case by case situation because the distinct activation functions may be indicative of a purposeful desire to give more weight to some of the neurons, and the percentage should respect this.
What you can do is to use a sigmoid transfer function on the output layer nodes (that accepts data ranges (-inf,inf) and outputs a value in [-1,1]).
Then by using the 1-of-n output encoding (one node for each class), you can map the range [-1,1] to [0,1] and use it as probability for each class value (note that this works naturally for more than just two classes).
The activation value of a single output neuron is a linearly weighted sum, and may be directly interpreted as an approximate probability if the network is trained to give outputs a range from 0 to 1. This would tend to be the case if the transfer function (or output function) in both the preceding stage and providing the final output is in the 0 to 1 range too (typically the sigmoidal logistic function). However, there is no guarantee that it will but repairs are possible. Moreover unless the sigmoids are logistic and the weights are constrained to be positive and sum to 1, it is unlikely. Generally a neural network will train in a more balanced way using the tanh sigmoid and weights and activations that range positive and negative (due to the symmetry of this model). Another factor is the prevalence of the class - if it is 50% then a 0.5 threshold is likely to be effective for logistic and a 0.0 threshold for tanh. The sigmoid is designed to push things towards the centre of the range (on backpropogation) and constrain it from going out of the range (in feedforward). The significance of the performance (with respect to the Bernoulli distribution) can also be interpreted as a probability that the neuron is making real predictions rather than guessing. Ideally the bias of the predictor to positives should match the prevalence of positives in the real world (which may vary at different times and places, e.g. bull vs bear markets, e.g. credit worthiness of people applying for loans vs people who fail to make loan payments) - calibrating to probabilities has the advantage that any desired bias can be set easily.
If you have two neurons for two classes, each can be interpreted independently as above, and the halved difference between them can also be. It is like flipping the negative class neuron and averaging. The differences can also give rise to a probability of significance estimate (using the T-test).
The Brier score and its Murphy decomposition give a more direct estimate of the probability that an average answer is correct, while Informedness gives the probability the classifier is making an informed decision rather than a guess, ROC AUC gives the probability a positive class will be ranked higher than a negative class (by a positive predictor), and Kappa will give a similar number that matches Informedness when prevalence = bias.
What you normally want is both a significance probability for the overall classifier (to ensure that you are playing on a real field, and not in an imaginary framework of guestimates) and a probability estimate for a specific example. There are various ways to calibrate, including doing a regression (linear or nonlinear) versus probability and using its inverse function to remap to a more accurate probability estimate. This can be seen by the Brier score improving, with the calibration component reducing towards 0, but the discrimination component remaining the same, as should ROC AUC and Informedness (Kappa is subject to bias and may worsen).
A simple non-linear way to calibrate to probabilities is to use the ROC curve - as the threshold changes for the output of a single neuron or the difference between two competing neurons, we plot the results true and false positive rates on a ROC curve (the false and true negative rates are naturally the complements, as what isn't really a positive is a negative). Then you scan the ROC curve (polyline) point by point (each time the gradient changes) sample by sample and the proportion of positive samples gives you a probability estimate for positives corresponding to the neural threshold that produced that point. Values between points on the curve can be linearly interpolated between those that are represented in the calibration set - and in fact any bad points in the ROC curve, represented by deconvexities (dents) can be smoothed over by the convex hull - probabilistically interpolating between the endpoints of the hull segment. Flach and Wu propose a technique that actually flips the segment, but this depends on information being used the wrong way round and although it could be used repeatedly for arbitrary improvement on the calibration set, it will be increasingly unlikely to generalize to a test situation.
(I came here looking for papers I'd seen ages ago on these ROC-based approaches - so this is from memory and without these lost references.)
I will be very prudent in interpreting the outputs of a neural networks (in fact any machine learning classifier) as a probability. The machine is trained to discriminate between classes, not to estimate the probability density. In fact, we don't have this information in the data, we have to infer it. For my experience I din't advice anyone to interpret directly the outputs as probabilities.
did you try prof. Hinton's suggestion of training the network with softmax activation function and cross entropy error?
as an example create a three layer network with the following:
linear neurons [ number of features ]
sigmoid neurons [ 3 x number of features ]
linear neurons [ number of classes ]
then train them with cross entropy error softmax transfer with your favourite optimizer stochastic descent/iprop plus/ grad descent. After training the output neurons should be normalized to sum of 1.
Please see http://en.wikipedia.org/wiki/Softmax_activation_function for details. Shark Machine Learning framework does provide Softmax feature through combining two models. And prof. Hinton an excellent online course # http://coursera.com regarding the details.
I can remember I saw an example of Neural network trained with back propagation to approximate the probability of an outcome in the book Introduction to the theory of neural computation (hertz krogh palmer). I think the key to the example was a special learning rule so that you didn't have to convert the output of a unit to probability, but instead you got automatically the probability as output.
If you have the opportunity, try to check that book.
(by the way, "boltzman machines", although less famous, are neural networks designed specifically to learn probability distributions, you may want to check them as well)
When using ANN for 2-class classification and logistic sigmoid activation function is used in the output layer, the output values could be interpreted as probabilities.
So if you choosing between 2 classes, you train using 1-of-C encoding, where 2 ANN outputs will have training values (1,0) and (0,1) for each of classes respectively.
To get probability of first class in percent, just multiply first ANN output to 100. To get probability of other class use the second output.
This could be generalized for multi-class classification using softmax activation function.
You can read more, including proofs of probabilistic interpretation here:
[1] Bishop, Christopher M. Neural networks for pattern recognition. Oxford university press, 1995.