I have a (time-expensive) operation that multiple metrics have in common. What would be the best way to share the operation result between the metrics avoiding the overhead of recalculating it each time?
You should create a special class to override tf.keras.callback.Callback() (thus implement your own callback class) and calculate the metrics that you need by overriding the method on_epoch_end().
Then, you can calculate some of your metrics, say, on the validation set, and you thus manually ensure that if you calculate for example TP + FP, you do use this sum to calculate the precision (TP / (TP + FP)) rather than recalculating it.
Manually doing so ensures no additional/superfluous computations are made.
Related
I use keras fit() method with custom metrics passed to model.
The metrics are stateful - i.e. are a subclass of a Metric, as described in https://keras.io/api/metrics/#as-subclasses-of-metric-stateful
When I run the code in a multi-gpu environment using a tf.distribute.MirroredStrategy() my metric code is called on every GPU separately with batch_size/no_of_gpus examples passed, which is reasonable to expect.
What happens next is that multiple scalars (one from every GPU) of the metric value need to be reduced to a single scalar, and what I get all the time is a sum reduction, while I would like to control that.
Keep in mind, that reduction parameter is the one of Loss in keras, and there is no such thing in the Metric class: https://github.com/tensorflow/tensorflow/blob/acbc065f8eb2ed05c7ab5c42b5c5bd6abdd2f91f/tensorflow/python/keras/metrics.py#L87
(the only crazy thing I tried was to inherit from a Mean class that is a subclass of a Metric but that didn't change anything)
reduction is mentioned in the metrics code, however this is a reduction over multiple accumulated values in a single metric object, and in multi-gpu setting - this is not the case, as every metric works in its own GPU and is somehow aggregated at the end.
The way I debugged it to understand this behaviour was - I was printing the shapes and the results inside update_state method of the metric. And then I looked at value of the metric in logs object in on_batch_end callback.
I tried looking at TF code, but couldn't find the place this is happening.
I would like to be able to control this behaviour - so either pick 'mean' or 'sum' for the metric, or at least know where it is being done in the code.
Edited: I guess this https://github.com/tensorflow/tensorflow/issues/39268 sheds some more light on this issue
I am facing the same problem as you (and that's why I found your question).
Seeing that it's been 15 days since you asked the question and there are no answers/comments yet, I thought I might share my temporary workaround.
Like you, I also think that a SUM reduction has been performed when combining progress over multiple GPUs. What I did is to pass the number of GPUs (e.g. given by the num_replicas_in_sync attribute of your tf.distribute strategy object) into the __init__(...) constructor of your sub-classed metric object, and use it to divide the return value in the results() method.
Potentially, you could also use tf.distribute.get_strategy() from within the metric object to make it "strategy aware", and use the information to decide how to modify the values in an ad hoc manner so that the SUM reduction will produce what you want.
I hope this helps for now, whether as a suggestion or as a confirmation that you're not alone on this.
When implementing the subclass of the Keras Metric class, you have to override the merge_state() function correctly. If you do not override this function, the default implementation will be used - which is a simple sum.
See: https://www.tensorflow.org/api_docs/python/tf/keras/metrics/Metric
I have been trying to implement this paper . Basically what I want to do is sum the per client loss and compare the same with previous epoch. Then for each constituent layer of the model compare the KL divergence between the weights of the server and the client model to get the layer specific parameter updates and then doing a softmax and to decide whether an adaptive update or a normal FedAvg approach is needed.
The algorithm is as follows-
FedMed
I tried to make use of the code here to build a custom federated avg process. I got the basic understanding that there are some tf.computations and some tff.computations which are involved. I get that I need to make changes in the orchestration logic in the run_one_round function and basically manipulate the client outputs to do adaptive averaging instead of the vanilla federated averaging. The client_update tf.computation function basically returns all the values that I need i.e the weights_delta (can be used for client based model weights), model_output(which can be used to calculate the loss).
But I am not sure where exactly I should make the changes.
#tff.federated_computation(federated_server_state_type,
federated_dataset_type)
def run_one_round(server_state, federated_dataset):
server_message = tff.federated_map(server_message_fn, server_state)
server_message_at_client = tff.federated_broadcast(server_message)
client_outputs = tff.federated_map(
client_update_fn, (federated_dataset, server_message_at_client))
weight_denom = client_outputs.client_weight
# todo
# instead of using tff.federated_mean I wish to do a adaptive aggregation based on the client_outputs.weights_delta and server_state model
round_model_delta = tff.federated_mean(
client_outputs.weights_delta, weight=weight_denom)
#client_outputs.weights_delta has all the client model weights.
#client_outputs.client_weight has the number of examples per client.
#client_outputs.model_output has the output of the model per client example.
I want to make use of the server model weights using server_state object.
I want to calculate the KL divergence between the weights of server model and each client's model per layer. Then use a relative weight to aggregate the client weights instead of vanilla federated averaging.
Instead of using tff.federated_mean I wish to use a different strategy basically an adaptive one based on the algorithm above.
So I needed some suggestions on how to go about implementing this.
Basically what I want to do is :
1)Sum all the values of client losses.
2)Calculate the KL divergence per layerbasis of all the clients with server and then determine whether to use adaptive optimization or FedAvg.
Also is there a way to manipulate this value as a python value which will be helpful for debugging purposes( I tried to use tf.print but that was not helpful either). Thanks!
Simplest option: compute weights for mean on clients
If I read the algorithm above correctly, we need only compute some weights for a mean on-the-fly. tff.federated_mean accepts an optional CLIENTS-placed weight argument, so probably the simplest option here is to compute the desired weights on the clients and pass them in to the mean.
This would look something like (assuming the appropriate definitions of the variables used below, which we will comment on):
#tff.federated_computation(...)
def round_function(...):
...
# We assume there is a tff.Computation training_fn that performs training,
# and we're calling it here on the correct arguments
trained_clients = tff.federated_map(training_fn, clients_placed_arguments)
# Next we assume there is a variable in-scope server_model,
# representing the 'current global model'.
global_model_at_clients = tff.federated_broadcast(server_model)
# Here we assume a function compute_kl_divergence, which takes
# two structures of tensors and computes the KL divergence
# (as a scalar) between them. The two arguments here are clients-placed,
# so the result will be as well.
kl_div_at_clients = tff.federated_map(compute_kl_divergence,
(global_model_at_clients, trained_clients))
# Perhaps we wish to not use raw KL divergence as the weight, but rather
# some function thereof; if so, we map a postprocessing function to
# the computed divergences. The result will still be clients-placed.
mean_weight = tff.federated_map(postprocess_divergence, kl_div_at_clients)
# Now we simply use the computed weights in the mean.
return tff.federated_mean(trained_clients, weight=mean_weight)
More flexible tool: tff.federated_reduce
TFF generally encourages algorithm developers to implement whatever they can 'in the aggregation', and as such exposes some highly customizable primitives like tff.federated_reduce, which allow you to run arbitrary TensorFlow "in the stream" between clients and server. If the above reading of the desired algorithm is incorrect and something more involved is needed, or you wish to flexibly experiment with totally different notions of aggregation (something TFF encourages and is designed to support), this may be the option for you.
In TFF's heuristic typing language, tff.federated_reduce has signature:
<{T}#CLIENTS, U, (<U, T> -> U)> -> U#SERVER
Meaning, federated_reduce take a value of type T placed at the clients, a 'zero' in a reduction algebra of type U, and a function accepting a U and a T and producing a U, and applies this function 'in the stream' on the way between clients and server, producing a U placed at the server. The function (<U, T> -> U) will be applied to the partially accumulated value U, and the 'next' element in the stream T (note however that TFF does not guarantee ordering of these values), returning another partially accumulated value U. The 'zero' should represent whatever 'partially accumulated' means over the empty set in your application; this will be the starting point of the reduction.
Application to this problem
The components
Your reduction function needs access to two pieces of data: the global model state and the result of training on a given client. This maps quite nicely to the type T. In this application, we will have something like:
T = <server_model=server_model_type, trained_model=trained_model_type>
These two types are likely to be the same, but may not necessarily be so.
Your reduction function will accept the partial aggregate, your server model and your client-trained model, returning a new partial aggregate. Here we will start assuming the same reading of the algorithm as above, that of a weighted mean with particular weights. Generally, the easiest way to compute a mean is to keep two accumulators, one for numerator and one for denominator. This will affect the choice of zero and reduction function below.
Your zero should contain a structure of tensors with value 0 mapping to the weights of your model--this will be the numerator. This would be generated for you if you had an aggregation like tff.federated_sum (as TFF knows what the zero should be), but for this case you'll have to get your hands on such a tensor yourself. This shouldn't be too hard with tf.nest.map_structure and tf.zeros_like.
For the denominator, we will assume we just need a scalar. TFF and TF are much more flexible than this--you could keep a per-layer or per-parameter denominator if desired--but for simplicity we will assume that we just want to divide by a single float in the end.
Therefore our type U will be something like:
U = <numerator=server_model_type, denominator=tf.float32>
Finally we come to our reduction function. It will be more or less a different composition of the same pieces above; we will make slightly tighter assumptions about them here (in particular, that all the local functions are tff.tf_computations--a technical assumption, arguably a bug on TFF). Our reduction function will be along the lines (assuming appropriate type aliases):
#tff.tf_computation(U, T)
def reduction(partial_accumulate, next_element):
kl_div = compute_kl_divergence(
next_element.server_model, next_element.trained_model)
weight = postprocess_divergence(kl_div)
new_numerator = partial_accumulate.numerator + weight * next_element.trained_model
new_denominator = partial_accumulate.denominator + weight
return collections.OrderedDict(
numerator=new_numerator, denominator=new_denominator)
Putting them together
The basic outline of a round will be similar to the above; but we have put more computation 'in the stream', and consequently there wil be less on the clients. We assume here the same variable definitions.
#tff.federated_computation(...)
def round_function(...):
...
trained_clients = tff.federated_map(training_fn, clients_placed_arguments)
global_model_at_clients = tff.federated_broadcast(server_model)
# This zip I believe is not necessary, but it helps my mental model.
reduction_arg = tff.federated_zip(
collections.OrderedDict(server_model=global_model_at_clients,
trained_model=trained_clients))
# We assume a zero as specified above
return tff.federated_reduce(reduction_arg,
zero,
reduction)
I have a multiclass model with 4 classes. I have already implemented a callback able to calculate the precision/recall for each class and their macro average. But for some technical reason, I have to calculate them using the metrics mechanism.
I'm using TensorFlow 2 and Keras 2.3.0. I have already used the tensorflow.keras.metrics.Recall/Precision to get the class-wise metrics:
metrics_list = ['accuracy']
metrics_list.extend([Recall(class_id=i, name="recall_{}".format(label_names[i])) for i in range(n_category)])
metrics_list.extend([Precision(class_id=i, name="precision_{}".format(label_names[i])) for i in range(n_category)])
model = Model(...)
model.compile(...metrics=metrics_list)
However, this solution is not satisfying:
firstly, tensorflow.keras.metrics.Recall/Precision uses a threshold to define the affiliation to a class, while it should use argmax to define the most probable class, if class_id is defined
Secondly, I have to create 2 new metrics that would calculate the average over all classes, which itself requires to calculate the class-wise metrics. This is inelegant and inefficient to calculate twice the same thing.
Is there a way to create a class or a function that would calculate directly the class-wise and the average predicion/recall using the TensorFlow/Keras metrics logic?
Apparently I can easily obtain the confusion matrix using tf.math.confusion_matrix(). However, I do not see how to inject a list of scalar at once, instead of returning a single scalar.
Any comment is welcomed!
It occurs that in my very specific case, I can simply use CategoricalAccuracy() as unique metric because i'm using a batch_size=1. It this case, accuracy=recall=precision={1.|0.} for a batch. That only partially solve the problem. The best solution would be to update the confusion matrix using argmax at each batch end, then calculate the Precision/Recall based on that. I don't known how it is possible to do that yet, but it should be doable.
× 19778
Extract mean from an ARMA(p,q) process
mean of an AR(p) process is mean= (intercept)/(1-phi1-phi2-..-phip) ,is this formula the same for an ARMA(p,q) also..?
ARMA(p,q):
Xt= intercept + Phi1 Xt-1 + Phi2 Xt-2 + ...+ Phip Xt-p + Epsilont+ theta1 Epsilont-1 +..+
thetaq Epsilont-q
where phi1,phi2,..,phip are the coefficients of AR part..
more info: I am comparing mean of a time series before an event and after an event..generally one can take a simple mean and compare. but since my data is auto-correlated I want to model the time series before and after so I can compare before and after means .. hence I need to extract the mean from a ARMA(p,q) process.
Thank You in advance
Yes, the moving average terms do not affect the marginal mean, they only affect the variance.
In comparing the two means, you will also need to adjust the variances because of the autocorrelations.
You are better off fitting an intervention model rather than trying to handle the autocorrelations by hand.
so I am making a simple simulation of different planets with individual velocity flying around space and orbiting each other.
I plan to simulate their pull on each other by considering each planet as projecting their own "gravity vector field." Each time step I'm going to add the vectors outputted from each planets individual vector field equation (V = -xj + (-yj) or some notation like it) except the one being effected in the calculation, and use the effected planets position as input to the equations.
However this would inaccurate, and does not consider the gravitational pull as continuous and constant. Bow do I calculate the movement of my planets if each is continuously effecting the others?
Thanks!
In addition to what Blender writes about using Newton's equations, you need to consider how you will be integrating over your "acceleration field" (as you call it in the comment to his answer).
The easiest way is to use Euler's Method. The problem with that is it rapidly diverges, but it has the advantage of being easy to code and to be reasonably fast.
If you are looking for better accuracy, and are willing to sacrifice some performance, one of the Runge-Kutta methods (probably RK4) would ordinarily be a good choice. I'll caution you that if your "acceleration field" is dynamic (i.e. it changes over time ... perhaps as a result of planets moving in their orbits) RK4 will be a challenge.
Update (Based on Comment / Question Below):
If you want to calculate the force vector Fi(tn) at some time step tn applied to a specific object i, then you need to compute the force contributed by all of the other objects within your simulation using the equation Blender references. That is for each object, i, you figure out how all of the other objects pull (apply force) and those vectors when summed will be the aggregate force vector applied to i. Algorithmically this looks something like:
for each object i
Fi(tn) = 0
for each object j ≠ i
Fi(tn) = Fi(tn) + G * mi * mj / |pi(tn)-pj(tn)|2
Where pi(tn) and pj(tn) are the positions of objects i and j at time tn respectively and the | | is the standard Euclidean (l2) normal ... i.e. the Euclidean distance between the two objects. Also, G is the gravitational constant.
Euler's Method breaks the simulation into discrete time slices. It looks at the current state and in the case of your example, considers all of the forces applied in aggregate to all of the objects within your simulation and then applies those forces as a constant over the period of the time slice. When using
ai(tn) = Fi(tn)/mi
(ai(tn) = acceleration vector at time tn applied to object i, Fi(tn) is the force vector applied to object i at time tn, and mi is the mass of object i), the force vector (and therefore the acceleration vector) is held constant for the duration of the time slice. In your case, if you really have another method of computing the acceleration, you won't need to compute the force, and can instead directly compute the acceleration. In either event, with the acceleration being held as constant, the position at time tn+1, p(tn+1) and velocity at time tn+1, v(tn+1), of the object will be given by:
pi(tn+1) = 0.5*ai(tn)*(tn+1-tn)2 + vi(tn)*(tn+1-tn)+pi(tn)
vi(tn+1) = ai(tn+1)*(tn+1-tn) + vi(tn)
The RK4 method fits the driver of your system to a 2nd degree polynomial which better approximates its behavior. The details are at the wikipedia site I referenced above, and there are a number of other resources you should be able to locate on the web. The basic idea is that instead of picking a single force value for a particular timeslice, you compute four force vectors at specific times and then fit the force vector to the 2nd degree polynomial. That's fine if your field of force vectors doesn't change between time slices. If you're using gravity to derive the vector field, and the objects which are the gravitational sources move, then you need to compute their positions at each of the four sub-intervals in order compute the force vectors. It can be done, but your performance is going to be quite a bit poorer than using Euler's method. On the plus side, you get more accurate motion of the objects relative to each other. So, it's a challenge in the sense that it's computationally expensive, and it's a bit of a pain to figure out where all the objects are supposed to be for your four samples during the time slice of your iteration.
There is no such thing as "continuous" when dealing with computers, so you'll have to approximate continuity with very small intervals of time.
That being said, why are you using a vector field? What's wrong with Newton?
And the sum of the forces on an object is that above equation. Equate the two and solve for a
So you'll just have to loop over all the objects one by one and find the acceleration on it.