How can change the shape of output of rnn network? - recurrent-neural-network

How can change the output of tf.nn.dynamic_rnn from [batch_size, max_stepsize, num_hidden] to [batch_size, num_hidden]?
self.outputs = tf.reshape(outputs, [-1, num_hidden]) gives [batch_size * max_stepsize, num_hidden] but I am looking for [batch_size , num_hidden] and I don’t want to use tf.reshape(outputs, [batch_size, num_hidden]).
I tried shape = tf.shape(outputs)[1] and then tf.squeeze(outputs, shape) but it gives error:
TypeError: Expected list for 'axis' argument to 'squeeze' Op, not <tf.Tensor 'lstm/strided_slice:0' shape=() dtype=int32>.
any solutions?

Related

Finding peaks with minimum peak width in R - similar to MATLAB function

I need to find peaks in a time series data, but the result needs to be equal to the result of the findpeaks function in MATLAB, with the argument 'MinPeakWidth" set to 10. I have already tried a lot of functions in order to achieve this: pracma::findpeaks, fluoR::find_peaks, splus2R::peaks, IDPmisc::peaks (this one has one argument regarding peak width, but the result is not the same). I have already looked in other functions as well, including packages for chromatography and spectoscropy analysis in bioconductor. Beyond that, I have tried the functions (and little alterations) from this other question in stackoverflow: Finding local maxima and minima
The findpeaks function in MATLAB is used for finding local maximas and has the following charcateristics:
Find the local maxima. The peaks are output in order of occurrence. The first sample is not included despite being the maximum. For the flat peak, the function returns only the point with lowest index.
The explanation for the "MinPeakWidth' argument in MATLAB web site is
Minimum peak width, specified as the comma-separated pair consisting of 'MinPeakWidth' and a positive real scalar. Use this argument to select only those peaks that have widths of at least 'MinPeakWidth'.
If you specify a location vector, x, then 'MinPeakWidth' must be expressed in terms of x. If x is a datetime array, then specify 'MinPeakWidth' as a duration scalar or as a numeric scalar expressed in days.
If you specify a sample rate, Fs, then 'MinPeakWidth' must be expressed in units of time.
If you specify neither x nor Fs, then 'MinPeakWidth' must be expressed in units of samples.
Data Types: double | single | duration
This is the data:
valores <- tibble::tibble(V1 = c(
0.04386573, 0.06169861, 0.03743560, 0.04512523, 0.04517977, 0.02927114, 0.04224937, 0.06596527, 2.15621006, 0.02547804, 0.03134409, 0.02867694,
0.08251871, 0.03252856, 0.06901365, 0.03201109, 0.04214851, 0.04679828, 0.04076178, 0.03922274, 1.65163662, 0.03630282, 0.04146608, 0.02618668,
0.04845364, 0.03202031, 0.03699149, 0.02811389, 0.03354410, 0.02975296, 0.03378896, 0.04440788, 0.46503730, 0.06128226, 0.01934736, 0.02055138,
0.04233819, 0.03398005, 0.02528630, 0.03694652, 0.02888223, 0.03463824, 0.04380172, 0.03297124, 0.04850558, 0.04579087, 1.48031231, 0.03735059,
0.04192204, 0.05789367, 0.03819694, 0.03344671, 0.05867103, 0.02590745, 0.05405133, 0.04941912, 0.63658824, 0.03134409, 0.04151859, 0.03502503,
0.02182294, 0.15397702, 0.02455722, 0.02775277, 0.04596132, 0.03900906, 0.03383408, 0.03517160, 0.02927114, 0.03888822, 0.03077891, 0.04236406,
0.05663730, 0.03619537, 0.04294887, 0.03497815, 0.03995837, 0.04374904, 0.03922274, 0.03596561, 0.03157820, 0.26390591, 0.06596527, 0.04050374,
0.02888223, 0.03824380, 0.05459656, 0.02969611, 0.86277224, 0.02385613, 0.03888451, 0.06496997, 0.03930725, 0.02931837, 0.06021005, 0.03330982,
0.02649659, 0.06600261, 0.02854480, 0.03691669, 0.06584168, 0.02076757, 0.02624355, 0.03679596, 0.03377049, 0.03590172, 0.03694652, 0.03575540,
0.02532416, 0.02818711, 0.04565318, 0.03252856, 0.04121822, 0.03147210, 0.05002047, 0.03809792, 0.02802299, 0.03399243, 0.03466543, 0.02829443,
0.03339476, 0.02129232, 0.03103367, 0.05071605, 0.03590172, 0.04386435, 0.03297124, 0.04323263, 0.03506247, 0.06225121, 0.02862442, 0.02862442,
0.06032925, 0.04400082, 0.03765090, 0.03477973, 0.02024540, 0.03564245, 0.05199116, 0.03699149, 0.03506247, 0.02129232, 0.02389752, 0.04996414,
0.04281258, 0.02587514, 0.03079668, 0.03895791, 0.02639014, 0.07333564, 0.02639014, 0.04074970, 0.04346211, 0.06032925, 0.03506247, 0.04950545,
0.04133673, 0.03835127, 0.02616212, 0.03399243, 0.02962473, 0.04800780, 0.03517160, 0.04105323, 0.03649472, 0.03000509, 0.05367187, 0.03858981,
0.03684529, 0.02941408, 0.04733265, 0.02590745, 0.02389752, 0.02385495, 0.03649472, 0.02508245, 0.02649659, 0.03152265, 0.02906310, 0.04950545,
0.03497815, 0.04374904, 0.03610649, 0.03799523, 0.02912771, 0.03694652, 0.05105353, 0.03000509, 0.02902378, 0.06425520, 0.05660319, 0.03065341,
0.04449069, 0.03638436, 0.02582273, 0.03753463, 0.02756006, 0.07215131, 0.02418869, 0.03431030, 0.04474425, 0.42589279, 0.02879489, 0.02872819,
0.02512494, 0.02450022, 0.03416346, 0.04560013, 1.40417366, 0.04784363, 0.04950545, 0.04685682, 0.03346052, 0.03255004, 0.07296053, 0.04491526,
0.02910482, 0.05448995, 0.01934736, 0.02195528, 0.03506247, 0.03157064, 0.03504810, 0.03754736, 0.03301058, 0.06886929, 0.03994190, 0.05130644,
0.21007323, 0.05630628, 0.02893721, 0.03683226, 0.03825290, 0.02494987, 0.02633410, 0.02721408, 0.03798986, 0.33473991, 0.04236406, 0.02389752,
0.03562747, 0.04662421, 0.02373767, 0.04918125, 0.04478894, 0.02418869, 0.03511514, 0.02871556, 0.05586166, 0.49014922, 0.03406339, 0.84823093,
0.03416346, 0.08729506, 0.03147210, 0.02889640, 0.06181828, 0.04940672, 0.03666858, 0.03019139, 0.03919279, 0.04864613, 0.03720420, 0.04726722,
0.04141298, 0.02862442, 0.29112744, 0.03964319, 0.05657445, 0.03930888, 0.04400082, 0.02722065, 0.03451685, 0.02911419, 0.02831578, 0.04001334,
0.05130644, 0.03134409, 0.03408579, 0.03232126, 0.03624218, 0.04708792, 0.06291741, 0.05663730, 0.03813209, 0.70582932, 0.04149421, 0.03607614,
0.03201109, 0.02055138, 0.03727305, 0.03182562, 0.02987404, 0.04142461, 0.03433624, 0.04264550, 0.02875086, 0.05797661, 0.04248705, 0.04476514))
From the data above, I obtain 22 peaks using pracma::findpeaks function with the code bellow:
picos_r <- pracma::findpeaks(-valores$V1, minpeakdistance = 10)
Using the MATLAB function
picos_matlab = findpeaks(-dado_r, 'MinPeakWidth', 10);
I obtain 11 peaks, as the following:
picos_matlab <- c(-0.02547804, -0.02618668, -0.01934736, -0.02182294, -0.0245572200000000, -0.0202454, -0.02385495, -0.01934736, -0.02373767, -0.02862442, -0.02722065)
I used pracma::findpeaks because it has already given an equal result in another part of the function that I am writting. I have already tried to change the code of the pracma::findpeaks, but with little success.

The package cardidates contains a heuristic peak hunting algorithm that can somewhat be fine-tuned using parameters xmax, minpeak and mincut. It was designed for a special problem, but may also used for other things. Here an example:
library("cardidates")
p <- peakwindow(valores$V1)
plot(p) # detects 14 peaks
p <- peakwindow(valores$V1, minpeak=0.18)
plot(p) # detects 11 peaks
Details are described in the package vignette and in https://doi.org/10.1007/s00442-007-0783-2
Another option is to run a smoother before peak detection.

I'm not sure what your test case is: -valores$V1, valores$V1, or -dado_r (what is that)?
I think pracma::findpeaks() does quite well if you do:
x <- valores$V1
P <- pracma::findpeaks(x,
minpeakdistance = 10, minpeakheight = sd(x))
plot(x, type = 'l', col = 4)
grid()
points(P[,2], P[, 1], pch=20, col = 2)
It finds 11 peaks that stick out while four or five others are too near to be counted. All the smaller ones (standard deviation) are being ignored.

Matix Exponential Layer (Custom: Keras in R)

I'm trying to make a layer in Keras (R) which (matrix) exponential a layer of shape (d,d).
ie.: Input to layer is a dxd matrix and the output is a dxd matrix which is the (matrix) exponential of the input matrix.
What I've Implemented to Date:
Here's what I've done (its a degree 4 approximation because I'm also not sure how to get the tensorflow matrix exponential command working in Keras):
# Matrix Exponential
Matrix_Exp<- R6::R6Class("KerasLayer",
inherit = KerasLayer,
public = list(
call = function(x, mask = NULL) {
# Initialize Tenor-like Object -> Tensor Objects
ord0 = k_eye((k_shape(x)[1]))
ord1 = x
ord2 = (1/2)*k_dot(x,x) # note x is square so this works
ord3 = (1/6)*k_dot(x,ord2)
ord4 = (1/24)*k_dot(x,ord3)
ord0+ord1+ord2 +ord3+ord4
},
compute_output_shape = function(input_shape) {
c(d,d)
}
)
)
# Create layer wrapper function
layer_Matrix_Exp <- function(object) {
create_layer(Matrix_Exp, object)
}
I'm plugging a model with this summary into the custom layer:
Model: "sequential_32"
_________________________________________________________________________________________________________________________________________________________________
Layer (type) Output Shape Param #
=================================================================================================================================================================
dense_63 (Dense) (None, 100) 400
_________________________________________________________________________________________________________________________________________________________________
dense_64 (Dense) (None, 4) 404
_________________________________________________________________________________________________________________________________________________________________
reshape_10 (Reshape) (None, 2, 2) 0
=================================================================================================================================================================
Total params: 804
Trainable params: 804
Non-trainable params: 0
_________________________________________________________________________________________________________________________________________________________________
Problem/Error:
But I run into this error when passing layers_NE %>% layer_Matrix_Exp
WARNING:tensorflow:Entity <function wrap_fn.<locals>.fn at 0x7fbdd0cf2b90> could not be transformed and will be executed as-is. Please report this to the AutoGraph team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output. Cause: Evaluation error: object 'size' not found.
Error in py_call_impl(callable, dots$args, dots$keywords) :
RuntimeError: in converted code:
/scratch/users/BIM/R/x86_64-redhat-linux-gnu-library/3.6/keras/python/kerastools/layer.py:30 call *
return self.r_call(inputs, mask)
<string>:4 fn
/scratch/users/BIM/R/x86_64-redhat-linux-gnu-library/3.6/reticulate/python/rpytools/call.py:21 python_function
raise RuntimeError(res[kErrorKey])
RuntimeError: Evaluation error: object 'size' not found.
Note:
The problem is coming from the identity part but I don't know how to fix this.
Question:
How to fix error.
How to replace the order 4 (manual) approximation to the matrix exponential with the keras equivalent to the tensorflow matrix exponential command.
Thanks in advance.

Unexpected symbol error in R that doesn't match my code

I am coding in R-studio and have a function called saveResults(). It takes:
sce - a Single Cell Experiment object.
opt - a list with five things
clusterLabels - simple dataframe with two columns
The important thing is that I receive an error stating:
Error: unexpected symbol in:
"saveResults(sce = sce, opt = opt, clusteInputs()
zhengMix"
which doesn't agree at all with the parameters I pass into the function. You can see this on the last line of the code block below: I pass in proper parameters, but I receive an error that says I have passed in clusteInputs(), and zhengMix instead of clusterLabels. I don't have a function called clusteInputs(), and zhengMix was several lines above.
# Save the clustering data
InstallAndLoadPackagesForSC3Clustering()
opt <- GetOptionInputs()
zhengMix <- FetchzhengMix(opt)
sce <- CreateSingleCellExperiment(zhengMix)
clusterLabels <- getClusterLabels(sce)
opt <- createNewDirectoriesToSaveData(opt)
saveResults <- function(sce, opt, clusterLabels){
print("Beginning process of saving results...")
maxClusters = ncol(clusterLabels)/2+1
for (n in 2:maxClusters){
savePCAasPDF(sce, opt, numOfClusters = n, clusterLabels)
saveClusterLabelsAsRDS(clusterLabels, numOfClusters = n, opt)
}
saveSilhouetteScores(sce, opt)
print("Done.")
}
saveResults(sce = sce, opt = opt, clusterLabels = clusterLabels)
Does anyone have an idea what is going on? I'm pretty stuck on this.

This isn't the best solution, but I fixed my own problem by removing the code out of the function and running it there caused no issues.

Data Buffer Exceeded error while connecting BW Cube

I am reading data from BW Cube with the following code
actual <- RSAPReadCube(conn, 'ZAPO_C17', chars=list('ZBEGINV', 'ZKF339', 'ZTOTJFF', 'ZOPENPL6', 'ZDLV_QTY', 'ZBACKLOG2'), kfigures=list('ZBEGINV', 'ZKF339', 'ZTOTJFF', 'ZOPENPL6', 'ZDLV_QTY', 'ZBACKLOG2'))
I am getting following error:
Error: Problem Invoking RFC (RSDPL_CUBE_DATA_READ): 5 / INHERITED_ERROR / ID:RS_EXCEPTION Type:E Number:000 Subfield access (offset = 0, length = 1-) to adata object of size 30 outside valid limits
Please help

Resolved with correct argument - which has Characteristic Technical Name
actual <- RSAPReadCube(conn, 'ZAPO_C17', chars=list('0MATERIAL', 'ZBWSDVERS'), kfigures=list('ZBEGINV', 'ZKF339', 'ZTOTJFF', 'ZOPENPL6', 'ZDLV_QTY', 'ZBACKLOG2'))

??? Error using ==> plot - conversion to double from sym is not possible

I'm having some problems with my code. Here it is:
lambdaz = 1.2;
n = 24;
mu = 0.00055e9;
lambda = sym('lambda','clear');
W = (((2.*mu)./n.^2)).*((lambda.^n)+(lambdaz.^n)+((lambda.^-n).*(lambdaz.^-n))-3);
dW_dlambda = diff(W, lambda);
W2=(((2.*mu)./n.^2).*(lambda.^n))+(((2.*mu)./n.^2).*(lambdaz.^n))+(((2.*mu)./n.^2).*((lambda.^-n).*(lambdaz.^-n)))-(3.*((2.*mu)./n.^2))
dW2_dlambda=diff(W2,lambda)
x=((((lambda.^2).*(lambdaz))-1).^-1).*(dW_dlambda);
x2=((((lambda.^2).*(lambdaz))-1).^-1).*(dW2_dlambda)
P2 = int(x2,lambda)
P=int(x,lambda);
P=(0:1000:26700)
plot(lambda,P)
When I try to plot lambda against P I get the "conversion to double from sym is not possible" error message. I'm not particularly fantastic at Matlab so any help would be gratefully received!

The plot function only works for numeric inputs. Both lambda and P are symbolic expressions (at least before overwrote P by setting it equal to a vector after the integration) that cannot be directly converted to to floating point. You get the same error if try to something like double(sym('exp(x)')). You have two options. The first is the ezplot function in the Symbolic Toolbox:
...
P = int(x,lambda);
ezplot(P,[-5 5]); % Plot's P from lambda = -5 to lambda = 5
Or you can use the subs function:
...
P = int(x,lambda);
lambda = -5:0.01:5;
plot(lambda,real(subs(P,'lambda',lambda)))
axis([lambda(1) lambda(end) -1e15 1e15])
I used real to suppress a warning for negative values of lambda.

Resources