So I have an array of N 1D-signals (e.g. time series) with same number of samples per signal (all in equal resolution) and I want to define a differentiable loss function to penalize asynchrony among them and therefore be zero if all N 1D signals will be equal to each other. I've been searching the literature to find something but haven't had luck yet.
Few remarks:
1 - since N (number of signals) could be quite large I can not afford to calculate Mean squared loss between every single pair which could grow combinatorialy large. also I'm not quite sure whether it would be optimal in any mathematical sense for the goal to achieve.
There are two naive loss functions that I could think of :
a) Total variation loss for each time sample across all signals (to force to reach ideally zero variation). the problem is here the weight needs to be very large to yield zero varion. masking any other loss term that is going to be added and also there is no inherent order among the N signals, which doesnt make it suitable to TV loss to begin with.
b) minimizing the sum of variance at each time point among all signals. however, choice of the reference of variance (aka mean) could be crucial I believe as just using the sample mean might not really yield the desired result, not quite sure.
Related
I am modeling a perceptual process in tensorflow. In the setup I am interested in, the modeled agent is playing a resource game: it has to choose 1 out of n resouces, by relying only on the label that a classifier gives to the resource. Each resource is an ordered pair of two reals. The classifier only sees the first real, but payoffs depend on the second. There is a function taking first to second.
Anyway, ideally I'd like to train the classifier in the following way:
In each run, the classifier give labels to n resources.
The agent then gets the payoff of the resource corresponding to the highest label in some predetermined ranking (say, A > B > C > D), and randomly in case of draw.
The loss is taken to be the normalized absolute difference between the payoff thus obtained and the maximum payoff in the set of resources. I.e., (Payoff_max - Payoff) / Payoff_max
For this to work, one needs to run inference n times, once for each resource, before calculating the loss. Is there a way to do this in tensorflow? If I am tackling the problem in the wrong way feel free to say so, too.
I don't have much knowledge in ML aspects of this, but from programming point of view, I can see doing it in two ways. One is by copying your model n times. All the copies can share the same variables. The output of all of these copies would go into some function that determines the the highest label. As long as this function is differentiable, variables are shared, and n is not too large, it should work. You would need to feed all n inputs together. Note that, backprop will run through each copy and update your weights n times. This is generally not a problem, but if it is, I heart about some fancy tricks one can do by using partial_run.
Another way is to use tf.while_loop. It is pretty clever - it stores activations from each run of the loop and can do backprop through them. The only tricky part should be to accumulate the inference results before feeding them to your loss. Take a look at TensorArray for this. This question can be helpful: Using TensorArrays in the context of a while_loop to accumulate values
I have a question about a reason why setting TensorFlow's variable with small stddev.
I guess many people do test MNIST test code from TensorFlow beginner's guide.
As following it, the first layer's weights are initiated by using truncated_normal with stddev 0.1.
And I guessed if setting it with more bigger value, then it would be the same result, which is exactly accurate.
But although increasing epoch count, it doesn't work.
Is there anybody know this reason?
original :
W_layer = tf.Variable(tf.truncated_normal([inp.get_shape()[1].value, size],stddev=0.1), name='w_'+name)
#result : (990, 0.93000001, 0.89719999)
modified :
W_layer = tf.Variable(tf.truncated_normal([inp.get_shape()[1].value, size],stddev=200), name='w_'+name)
#result : (99990, 0.1, 0.098000005)
The reason is because you want to keep all the layer's variances (or standard deviations) approximately the same, and sane. It has to do with the error backpropagation step of the learning process and the activation functions used.
In order to learn the network's weights, the backpropagation step requires knowledge of the network's gradient, a measure of how strong each weight influences the input to reach the final output; layer's weight variance directly influences the propagation of gradients.
Say, for example, that the activation function is sigmoidal (e.g. tf.nn.sigmoid or tf.nn.tanh); this implies that all input values are squashed into a fixed output value range. For the sigmoid, it is the range 0..1, where essentially all values z greater or smaller than +/- 4 are very close to one (for z > 4) or zero (for z < -4) and only values within that range tend to have some meaningful "change".
Now the difference between the values sigmoid(5) and sigmoid(1000) is barely noticeable. Because of that, all very large or very small values will optimize very slowly, since their influence on the result y = sigmoid(W*x+b) is extremely small. Now the pre-activation value z = W*x+b (where x is the input) depends on the actual input x and the current weights W. If either of them is large, e.g. by initializing the weights with a high variance (i.e. standard deviation), the result will necessarily be (relatively) large, leading to said problem. This is also the reason why truncated_normal is used rather than a correct normal distribution: The latter only guarantees that most of the values are very close to the mean, with some less than 5% chance that this is not the case, while truncated_normal simply clips away every value that is too big or too small, guaranteeing that all weights are in the same range, while still being normally distributed.
To make matters worse, in a typical neural network - especially in deep learning - each network layer is followed by one or many others. If in each layer the output value range is big, the gradients will get bigger and bigger as well; this is known as the exploding gradients problem (a variation of the vanishing gradients, where gradients are getting smaller).
The reason that this is a problem is because learning starts at the very last layer and each weight is adjusted depending on how much it contributed to the error. If the gradients are indeed getting very big towards the end, the very last layer is the first one to pay a high toll for this: Its weights get adjusted very strongly - likely overcorrecting the actual problem - and then only the "remaining" error gets propagated further back, or up, the network. Here, since the last layer was already "fixed a lot" regarding the measured error, only smaller adjustments will be made. This may lead to the problem that the first layers are corrected only by a tiny bit or not at all, effectively preventing all learning there. The same basically happens if the learning rate is too big.
Finding the best weight initialization is a topic by itself and there are somewhat more sophisticated methods such as Xavier initialization or Layer-sequential unit variance, however small normally distributed values are usually simply a good guess.
Minimally, I would like to know how to achieve what is stated in the title. Specifically, signal.lfilter seems like the only implementation of a difference equation filter in scipy, but it is 1D, as shown in the docs. I would like to know how to implement a 2D version as described by this difference equation. If that's as simple as "bro, use this function," please let me know, pardon my naiveté, and feel free to disregard the rest of the post.
I am new to DSP and acknowledging there might be a different approach to answering my question so I will explain the broader goal and give context for the question in the hopes someone knows how do want I want with Scipy, or perhaps a better way than what I explicitly asked for.
To get straight into it, broadly speaking I am using vectorized computation methods (Numpy/Scipy) to implement a Monte Carlo simulation to improve upon a naive for loop. I have successfully abstracted most of my operations to array computation / linear algebra, but a few specific ones (recursive computations) have eluded my intuition and I continually end up in the digital signal processing world when I go looking for how this type of thing has been done by others (that or machine learning but those "frameworks" are much opinionated). The reason most of my google searches end up on scipy.signal or scipy.ndimage library references is clear to me at this point, and subsequent to accepting the "signal" representation of my data, I have spent a considerable amount of time (about as much as reasonable for a field that is not my own) ramping up the learning curve to try and figure out what I need from these libraries.
My simulation entails updating a vector of data representing the state of a system each period for n periods, and then repeating that whole process a "Monte Carlo" amount of times. The updates in each of n periods are inherently recursive as the next depends on the state of the prior. It can be characterized as a difference equation as linked above. Additionally this vector is theoretically indexed on an grid of points with uneven stepsize. Here is an example vector y and its theoretical grid t:
y = np.r_[0.0024, 0.004, 0.0058, 0.0083, 0.0099, 0.0133, 0.0164]
t = np.r_[0.25, 0.5, 1, 2, 5, 10, 20]
I need to iteratively perform numerous operations to y for each of n "updates." Specifically, I am computing the curvature along the curve y(t) using finite difference approximations and using the result at each point to adjust the corresponding y(t) prior to the next update. In a loop this amounts to inplace variable reassignment with the desired update in each iteration.
y += some_function(y)
Not only does this seem inefficient, but vectorizing things seems intuitive given y is a vector to begin with. Furthermore I am interested in preserving each "updated" y(t) along the n updates, which would require a data structure of dimensions len(y) x n. At this point, why not perform the updates inplace in the array? This is wherein lies the question. Many of the update operations I have succesfully vectorized the "Numpy way" (such as adding random variates to each point), but some appear overly complex in the array world.
Specifically, as mentioned above the one involving computing curvature at each element using its neighbouring two elements, and then imediately using that result to update the next row of the array before performing its own curvature "update." I was able to implement a non-recursive version (each row fails to consider its "updated self" from the prior row) of the curvature operation using ndimage generic_filter. Given the uneven grid, I have unique coefficients (kernel weights) for each triplet in the kernel footprint (instead of always using [1,-2,1] for y'' if I had a uniform grid). This last part has already forced me to use a spatial filter from ndimage rather than a 1d convolution. I'll point out, something conceptually similar was discussed in this math.exchange post, and it seems to me only the third response saliently addressed the difference between mathematical notion of "convolution" which should be associative from general spatial filtering kernels that would require two sequential filtering operations or a cleverly merged kernel.
In any case this does not seem to actually address my concern as it is not about 2D recursion filtering but rather having a backwards looking kernel footprint. Additionally, I think I've concluded it is not applicable in that this only allows for "recursion" (backward looking kernel footprints in the spatial filtering world) in a manner directly proportional to the size of the recursion. Meaning if I wanted to filter each of n rows incorporating calculations on all prior rows, it would require a convolution kernel far too big (for my n anyways). If I'm understanding all this correctly, a recursive linear filter is algorithmically more efficient in that it returns (for use in computation) the result of itself applied over the previous n samples (up to a level where the stability of the algorithm is affected) using another companion vector (z). In my case, I would only need to look back one step at output signal y[n-1] to compute y[n] from curvature at x[n] as the rest works itself out like a cumsum. signal.lfilter works for this, but I can't used that to compute curvature, as that requires a kernel footprint that can "see" at least its left and right neighbors (pixels), which is how I ended up using generic_filter.
It seems to me I should be able to do both simultaneously with one filter namely spatial and recursive filtering; or somehow I've missed the maths of how this could be mathematically simplified/combined (convolution of multiples kernels?).
It seems like this should be a common problem, but perhaps it is rarely relevant to do both at once in signal processing and image filtering. Perhaps this is why you don't use signals libraries solely to implement a fast monte carlo simulation; though it seems less esoteric than using a tensor math library to implement a recursive neural network scan ... which I'm attempting to do right now.
EDIT: For those familiar with the theoretical side of DSP, I know that what I am describing, the process of designing a recursive filters with arbitrary impulse responses, is achieved by employing a mathematical technique called the z-transform which I understand is generally used for two things:
converting between the recursion coefficients and the frequency response
combining cascaded and parallel stages into a single filter
Both are exactly what I am trying to accomplish.
Also, reworded title away from FIR / IIR because those imply specific definitions of "recursion" and may be confusing / misnomer.
All,
I am doing Bayesian modeling using rjags. However, when the number of observation is larger than 1000. The graph size is too big.
More specifically, I am doing a Bayesian ranking problem. Traditionally, one observation means one X[i, 1:N]-Y[i] pair, where X[i, 1:N] means the i-th item is represented by a N-size predictor vector, and Y[i] is a response. The objective is to minimize the point-wise error of predicted values,for example, least square error.
A ranking problem is different. Since we more care about the order, we use a pair-wise 1-0 indicator to represent the order between Y[i] and Y[j], for example, when Y[i]>Y[j], I(i,j)=1; otherwise I(i,j)=0. We treat this 1-0 indicator as an observation. Therefore, assuming we have K items: Y[1:K], the number of indicator is 0.5*K*(K-1). Hence when K is increased from 500 to 5000, the number of observations is very large, i.e. from 500^2 to 5000^2. The garph size of the rjags model is large too, for example graph size > 500,000. And the log-posterior will be very small.
And it takes a long time to complete the training. I think the consumed time is >40 hours. It is not practical for me to do further experiment. Therefore, do you have any idea to speed up the rjags. I heard that the RStan is faster than Rjags. Any one who has similar experience?
I am using simulated annealing to solve an NP-complete resource scheduling problem. For each candidate ordering of the tasks I compute several different costs (or energy values). Some examples are (though the specifics are probably irrelevant to the question):
global_finish_time: The total number of days that the schedule spans.
split_cost: The number of days by which each task is delayed due to interruptions by other tasks (this is meant to discourage interruption of a task once it has started).
deadline_cost: The sum of the squared number of days by which each missed deadline is overdue.
The traditional acceptance probability function looks like this (in Python):
def acceptance_probability(old_cost, new_cost, temperature):
if new_cost < old_cost:
return 1.0
else:
return math.exp((old_cost - new_cost) / temperature)
So far I have combined my first two costs into one by simply adding them, so that I can feed the result into acceptance_probability. But what I would really want is for deadline_cost to always take precedence over global_finish_time, and for global_finish_time to take precedence over split_cost.
So my question to Stack Overflow is: how can I design an acceptance probability function that takes multiple energies into account but always considers the first energy to be more important than the second energy, and so on? In other words, I would like to pass in old_cost and new_cost as tuples of several costs and return a sensible value .
Edit: After a few days of experimenting with the proposed solutions I have concluded that the only way that works well enough for me is Mike Dunlavey's suggestion, even though this creates many other difficulties with cost components that have different units. I am practically forced to compare apples with oranges.
So, I put some effort into "normalizing" the values. First, deadline_cost is a sum of squares, so it grows exponentially while the other components grow linearly. To address this I use the square root to get a similar growth rate. Second, I developed a function that computes a linear combination of the costs, but auto-adjusts the coefficients according to the highest cost component seen so far.
For example, if the tuple of highest costs is (A, B, C) and the input cost vector is (x, y, z), the linear combination is BCx + Cy + z. That way, no matter how high z gets it will never be more important than an x value of 1.
This creates "jaggies" in the cost function as new maximum costs are discovered. For example, if C goes up then BCx and Cy will both be higher for a given (x, y, z) input and so will differences between costs. A higher cost difference means that the acceptance probability will drop, as if the temperature was suddenly lowered an extra step. In practice though this is not a problem because the maximum costs are updated only a few times in the beginning and do not change later. I believe this could even be theoretically proven to converge to a correct result since we know that the cost will converge toward a lower value.
One thing that still has me somewhat confused is what happens when the maximum costs are 1.0 and lower, say 0.5. With a maximum vector of (0.5, 0.5, 0.5) this would give the linear combination 0.5*0.5*x + 0.5*y + z, i.e. the order of precedence is suddenly reversed. I suppose the best way to deal with it is to use the maximum vector to scale all values to given ranges, so that the coefficients can always be the same (say, 100x + 10y + z). But I haven't tried that yet.
mbeckish is right.
Could you make a linear combination of the different energies, and adjust the coefficients?
Possibly log-transforming them in and out?
I've done some MCMC using Metropolis-Hastings. In that case I'm defining the (non-normalized) log-likelihood of a particular state (given its priors), and I find that a way to clarify my thinking about what I want.
I would take a hint from multi-objective evolutionary algorithm (MOEA) and have it transition if all of the objectives simultaneously pass with the acceptance_probability function you gave. This will have the effect of exploring the Pareto front much like the standard simulated annealing explores plateaus of same-energy solutions.
However, this does give up on the idea of having the first one take priority.
You will probably have to tweak your parameters, such as giving it a higher initial temperature.
I would consider something along the lines of:
If (new deadline_cost > old deadline_cost)
return (calculate probability)
else if (new global finish time > old global finish time)
return (calculate probability)
else if (new split cost > old split cost)
return (calculate probability)
else
return (1.0)
Of course each of the three places you calculate the probability could use a different function.
It depends on what you mean by "takes precedence".
For example, what if the deadline_cost goes down by 0.001, but the global_finish_time cost goes up by 10000? Do you return 1.0, because the deadline_cost decreased, and that takes precedence over anything else?
This seems like it is a judgment call that only you can make, unless you can provide enough background information on the project so that others can suggest their own informed judgment call.