Is there an easy way to calculate relative gradient error in tensorflow? All what is available is tf.test.compute_gradient_error but it computes absolute gradient error and not relative error. Of courser there're methods which compute numeric and theoretical jacobians but they are private.
I found the answer myself tf.test.compute_gradient returns two jacobians, so I can use them to find the relative gradient error. I.e. if I use L-infinity norm, I can take tf.test.compute_gradient_error and divide in on the maximum element of both jacobians.
Related
I have a dataframe which contains three more or less significant correlations between target column and other columns ( LinarRegressionModel.coef_ from sklearn shows 57, 97 and 79). And I don't know what exact model to choose: should I use only most correlated column for regression or use regression with all three predictors. Is there any way to compare models effectiveness? Sorry, I'm very new to data analysis, I couldn't google any tools for this task
Well first at all, you must know that when we are choosing the best model to apply to new data, we are going to choose the best model to fit out of sample data, which is the kind of samples that might are not present in the training process, after all, you want to predict new probabilities or cases. In your case, predict a new number.
So, how can we do this? Well, the best is to use metrics which can help us to choose which model is better for our dataset.
There are so many kinds of metrics for regression:
MAE: Mean absolute error is the mean of the absolute value of the errors. This is the easiest of the metrics to understand since it’s just the average error.
MSE: Mean squared error is the mean of the squared error. It’s more popular than a mean absolute error because the focus is geared more towards large errors.
RMSE: Root means the squared error is the square root of the mean squared error. This is one of the most popular of the evaluation metrics because root means the squared error is interpretable in the same units as the response vector or y units, making it easy to relate its information.
RAE: Relative absolute error, also known as the residual sum of a square, where y bar is a mean value of y, takes the total absolute error and normalizes it by dividing by the total absolute error of the simple predictor.
You can work with any of these, but I highly recommend to use MSE and RMSE.
I was playing around with Tensorflow creating a customized loss function and this question about general machine learning arose to my head.
My understanding is that the optimization algorithm needs a derivable cost function to find/approach a minimum, however we can use functions that are non-derivable such as the absolute function (there is no derivative when x=0). A more extreme example, I defined my cost function like this:
def customLossFun(x,y):
return tf.sign(x)
and I expected an error when running the code, but it actually worked (it didn't learn anything but it didn't crash).
Am I missing something?
You're missing the fact that the gradient of the sign function is somewhere manually defined in the Tensorflow source code.
As you can see here:
def _SignGrad(op, _):
"""Returns 0."""
x = op.inputs[0]
return array_ops.zeros(array_ops.shape(x), dtype=x.dtype)
the gradient of tf.sign is defined to be always zero. This, of course, is the gradient where the derivate exists, hence everywhere but not in zero.
The tensorflow authors decided to do not check if the input is zero and throw an exception in that specific case
In order to prevent TensorFlow from throwing an error, the only real requirement is that you cost function evaluates to a number for any value of your input variables. From a purely "will it run" perspective, it doesn't know/care about the form of the function its trying to minimize.
In order for your cost function to provide you a meaningful result when TensorFlow uses it to train a model, it additionally needs to 1) get smaller as your model does better and 2) be bounded from below (i.e. it can't go to negative infinity). It's not generally necessary for it to be smooth (e.g. abs(x) has a kink where the sign flips). Tensorflow is always able to compute gradients at any location using automatic differentiation (https://en.wikipedia.org/wiki/Automatic_differentiation, https://www.tensorflow.org/versions/r0.12/api_docs/python/train/gradient_computation).
Of course, those gradients are of more use if you've chose a meaningful cost function isn't isn't too flat.
Ideally, the cost function needs to be smooth everywhere to apply gradient based optimization methods (SGD, Momentum, Adam, etc). But nothing's going to crash if it's not, you can just have issues with convergence to a local minimum.
When the function is non-differentiable at a certain point x, it's possible to get large oscillations if the neural network converges to this x. E.g., if the loss function is tf.abs(x), it's possible that the network weights are mostly positive, so the inference x > 0 at all times, so the network won't notice tf.abs. However, it's more likely that x will bounce around 0, so that the gradient is arbitrarily positive and negative. If the learning rate is not decaying, the optimization won't converge to the local minimum, but will bound around it.
In your particular case, the gradient is zero all the time, so nothing's going to change at all.
If it didn't learn anything, what have you gained ? Your loss function is differentiable almost everywhere but it is flat almost anywhere so the minimizer can't figure out the direction towards the minimum.
If you start out with a positive value, it will most likely be stuck at a random value on the positive side even though the minima on the left side are better (have a lower value).
Tensorflow can be used to do calculations in general and it provides a mechanism to automatically find the derivative of a given expression and can do so across different compute platforms (CPU, GPU) and distributed over multiple GPUs and servers if needed.
But what you implement in Tensorflow does not necessarily have to be a goal function to be minimized. You could use it e.g. to throw random numbers and perform Monte Carlo integration of a given function.
I am using this function of tensorflow to get my function jacobian. Came across two problems:
The tensorflow documentation is contradicted to itself in the following two paragraph if I am not mistaken:
gradients() adds ops to the graph to output the partial derivatives of ys with respect to xs. It returns a list of Tensor of length len(xs) where each tensor is the sum(dy/dx) for y in ys.
Blockquote
Blockquote
Returns:
A list of sum(dy/dx) for each x in xs.
Blockquote
According to my test, it is, in fact, return a vector of len(ys) which is the sum(dy/dx) for each x in xs.
I do not understand why they designed it in a way that the return is the sum of the columns(or row, depending on how you define your Jacobian).
How can I really get the Jacobian?
4.In the loss, I need the partial derivative of my function with respect to input (x), but when I am optimizing with respect to the network weights, I define x as a placeholder whose value is fed later, and weights are variable, in this case, can I still define the symbolic derivative of function with respect to input (x)? and put it in the loss? ( which later when we optimize with respect to weights will bring second order derivative of the function.)
I think you are right and there is a typo there, it was probably meant to be "of length len(ys)".
For efficiency. I can't explain exactly the reasoning, but this seems to be a pretty fundamental characteristic of how TensorFlow handles automatic differentiation. See issue #675.
There is no straightforward way to get the Jacobian matrix in TensorFlow. Take a look at this answer and again issue #675. Basically, you need one call to tf.gradients per column/row.
Yes, of course. You can compute whatever gradients you want, there is no real difference between a placeholder and any other operation really. There are a few operations that do not have a gradient because it is not well defined or not implemented (in which case it will generally return 0), but that's all.
I use the scipy.optimize.minimize ( https://docs.scipy.org/doc/scipy/reference/tutorial/optimize.html ) function with method='L-BFGS-B.
An example of what it returns is here above:
fun: 32.372210618549758
hess_inv: <6x6 LbfgsInvHessProduct with dtype=float64>
jac: array([ -2.14583906e-04, 4.09272616e-04, -2.55795385e-05,
3.76587650e-05, 1.49213975e-04, -8.38440428e-05])
message: 'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'
nfev: 420
nit: 51
status: 0
success: True
x: array([ 0.75739412, -0.0927572 , 0.11986434, 1.19911266, 0.27866406,
-0.03825225])
The x value correctly contains the fitted parameters. How do I compute the errors associated to those parameters?
TL;DR: You can actually place an upper bound on how precisely the minimization routine has found the optimal values of your parameters. See the snippet at the end of this answer that shows how to do it directly, without resorting to calling additional minimization routines.
The documentation for this method says
The iteration stops when (f^k - f^{k+1})/max{|f^k|,|f^{k+1}|,1} <= ftol.
Roughly speaking, the minimization stops when the value of the function f that you're minimizing is minimized to within ftol of the optimum. (This is a relative error if f is greater than 1, and absolute otherwise; for simplicity I'll assume it's an absolute error.) In more standard language, you'll probably think of your function f as a chi-squared value. So this roughly suggests that you would expect
Of course, just the fact that you're applying a minimization routine like this assumes that your function is well behaved, in the sense that it's reasonably smooth and the optimum being found is well approximated near the optimum by a quadratic function of the parameters xi:
where Δxi is the difference between the found value of parameter xi and its optimal value, and Hij is the Hessian matrix. A little (surprisingly nontrivial) linear algebra gets you to a pretty standard result for an estimate of the uncertainty in any quantity X that's a function of your parameters xi:
which lets us write
That's the most useful formula in general, but for the specific question here, we just have X = xi, so this simplifies to
Finally, to be totally explicit, let's say you've stored the optimization result in a variable called res. The inverse Hessian is available as res.hess_inv, which is a function that takes a vector and returns the product of the inverse Hessian with that vector. So, for example, we can display the optimized parameters along with the uncertainty estimates with a snippet like this:
ftol = 2.220446049250313e-09
tmp_i = np.zeros(len(res.x))
for i in range(len(res.x)):
tmp_i[i] = 1.0
hess_inv_i = res.hess_inv(tmp_i)[i]
uncertainty_i = np.sqrt(max(1, abs(res.fun)) * ftol * hess_inv_i)
tmp_i[i] = 0.0
print('x^{0} = {1:12.4e} ± {2:.1e}'.format(i, res.x[i], uncertainty_i))
Note that I've incorporated the max behavior from the documentation, assuming that f^k and f^{k+1} are basically just the same as the final output value, res.fun, which really ought to be a good approximation. Also, for small problems, you can just use np.diag(res.hess_inv.todense()) to get the full inverse and extract the diagonal all at once. But for large numbers of variables, I've found that to be a much slower option. Finally, I've added the default value of ftol, but if you change it in an argument to minimize, you would obviously need to change it here.
One approach to this common problem is to use scipy.optimize.leastsq after using minimize with 'L-BFGS-B' starting from the solution found with 'L-BFGS-B'. That is, leastsq will (normally) include and estimate of the 1-sigma errors as well as the solution.
Of course, that approach makes several assumption, including that leastsq can be used and may be appropriate for solving the problem. From a practical view, this requires the objective function return an array of residual values with at least as many elements as variables, not a cost function.
You may find lmfit (https://lmfit.github.io/lmfit-py/) useful here: It supports both 'L-BFGS-B' and 'leastsq' and gives a uniform wrapper around these and other minimization methods, so that you can use the same objective function for both methods (and specify how to convert the residual array into the cost function). In addition, parameter bounds can be used for both methods. This makes it very easy to first do a fit with 'L-BFGS-B' and then with 'leastsq', using the values from 'L-BFGS-B' as starting values.
Lmfit also provides methods to more explicitly explore confidence limits on parameter values in more detail, in case you suspect the simple but fast approach used by leastsq might be insufficient.
It really depends what you mean by "errors". There is no general answer to your question, because it depends on what you're fitting and what assumptions you're making.
The easiest case is one of the most common: when the function you are minimizing is a negative log-likelihood. In that case the inverse of the hessian matrix returned by the fit (hess_inv) is the covariance matrix describing the Gaussian approximation to the maximum likelihood.The parameter errors are the square root of the diagonal elements of the covariance matrix.
Beware that if you are fitting a different kind of function or are making different assumptions, then that doesn't apply.
When I fitted a histogram using Gaussian its error on mean and sigma seems to be fine. You can look at it here .
But, when I first normalized the histogram and fitted it with Gaussian its parameters value is exactly same as previous case but the error on mean and sigma is almost equal to the actual value or greater.
One of the reason for this is that it may be happening because it is taking error as 1/sqrt{n} and after normalizing n decreased and hence error increased.
Please let me know what is happening and how I can fix it?
You probably want to call
hist->Sumw2()
before rescaling the histogram. Otherwise the uncertainties on all bin contents are just the square roots of the bin contents (which is a huge relative error for bin contents smaller than 1, which is the case when after rescaling). SumW2 triggers to store the sum of all weights squared and not only the bin contents (i.e. the sum of weights in each bin).
See also the documentation of Sumw2() for further details (and also the explanation of weights on the top of the TH1 documentation page).