Here is a part of my tensorflow RNN network code written in jupyter. The whole code runs perfect for the first time, however, running it furthermore produces an error. The code:
x = tf.placeholder(tf.float32)
y = tf.placeholder(tf.float32)
def recurrent_nn_model(x):
x = tf.transpose(x, [1,0,2])
x = tf.reshape(x, [-1, chunk_size])
x = tf.split(x, n_chunks, 0)
lstm_layer = {'hidden_state': tf.zeros([n_batches, lstm_size]),
'current_state': tf.zeros([n_batches, lstm_size])}
layer = {'weights': tf.Variable(tf.random_normal([lstm_size, n_classes])),
'biases': tf.Variable(tf.random_normal([n_classes]))}
lstm = rnn_cell.BasicLSTMCell(lstm_size)
rnn_outputs, rnn_states = rnn.static_rnn(lstm, x, dtype=tf.float32)
output = tf.add(tf.matmul(rnn_outputs[-1], layer['weights']),
layer['biases'])
return output
The error is:
Variable rnn/basic_lstm_cell/kernel already exists, disallowed. Did
you mean to set reuse=True in VarScope? Originally defined at:
If recurrent_nn_model is the whole network, just add this line to reset previously defined graph:
tf.reset_default_graph()
If you're intentionally calling recurrent_nn_model several times and combine these RNNs into one graph, you should use different variable scopes for each one:
with tf.variable_scope('lstm1'):
recurrent_nn_model(x1)
with tf.variable_scope('lstm2'):
recurrent_nn_model(x2)
Related
I am trying to use reinforcement learning in julia to teach a car that is constantly being accelerated backwards (but with a positive initial velocity) to apply brakes so that it gets as close to a target distance as possible before moving backwards.
To do this, I am making use of POMDPs.jl and crux.jl which has many solvers (I'm using DQN). I will list what I believe to be the relevant parts of the script first, and then more of it towards the end.
To define the MDP, I set the initial position, velocity, and force from the brakes as a uniform distribution over some values.
#with_kw struct SliderMDP <: MDP{Array{Float32}, Array{Float32}}
x0 = Distributions.Uniform(0., 80.)# Distribution to sample initial position
v0 = Distributions.Uniform(0., 25.) # Distribution to sample initial velocity
d0 = Distributions.Uniform(0., 2.) # Distribution to sample brake force
...
end
My state holds the values of (position, velocity, brake force), and the initial state is given as:
function POMDPs.initialstate(mdp::SliderMDP)
ImplicitDistribution((rng) -> Float32.([rand(rng, mdp.x0), rand(rng, mdp.v0), rand(rng, mdp.d0)]))
end
Then, I set up my DQN solver using crux.jl and called a function to solve for the policy
solver_dqn = DQN(π=Q_network(), S=s, N=30000)
policy_dqn = solve(solver_dqn, mdp)
calling solve() gives me the error MethodError: no method matching logpdf(::Distributions.Categorical{Float64, Vector{Float64}}, ::Nothing). I am quite sure that this comes from the initial state sampling, but I am not sure why or how to fix it. I have only been learning RL from various books and online lectures for a very short time, so any help regarding the error or my the model I set up (or anything else I'm oblivious to) would be appreciated.
More comprehensive code:
Packages:
using POMDPs
using POMDPModelTools
using POMDPPolicies
using POMDPSimulators
using Parameters
using Random
using Crux
using Flux
using Distributions
Rest of it:
#with_kw struct SliderMDP <: MDP{Array{Float32}, Array{Float32}}
x0 = Distributions.Uniform(0., 80.)# Distribution to sample initial position
v0 = Distributions.Uniform(0., 25.) # Distribution to sample initial velocity
d0 = Distributions.Uniform(0., 2.) # Distribution to sample brake force
m::Float64 = 1.
tension::Float64 = 3.
dmax::Float64 = 2.
target::Float64 = 80.
dt::Float64 = .05
γ::Float32 = 1.
actions::Vector{Float64} = [-.1, 0., .1]
end
function POMDPs.gen(env::SliderMDP, s, a, rng::AbstractRNG = Random.GLOBAL_RNG)
x, ẋ, d = s
if x >= env.target
a = .1
end
if d+a >= env.dmax || d+a <= 0
a = 0.
end
force = (d + env.tension) * -1
ẍ = force/env.m
# Simulation
x_ = x + env.dt * ẋ
ẋ_ = ẋ + env.dt * ẍ
d_ = d + a
sp = vcat(x_, ẋ_, d_)
reward = abs(env.target - x) * -1
return (sp=sp, r=reward)
end
function POMDPs.initialstate(mdp::SliderMDP)
ImplicitDistribution((rng) -> Float32.([rand(rng, mdp.x0), rand(rng, mdp.v0), rand(rng, mdp.d0)]))
end
POMDPs.isterminal(mdp::SliderMDP, s) = s[2] <= 0
POMDPs.discount(mdp::SliderMDP) = mdp.γ
mdp = SliderMDP();
s = state_space(mdp); # Using Crux.jl
function Q_network()
layer1 = Dense(3, 64, relu)
layer2 = Dense(64, 64, relu)
layer3 = Dense(64, length(3))
return DiscreteNetwork(Chain(layer1, layer2, layer3), [-.1, 0, .1])
end
solver_dqn = DQN(π=Q_network(), S=s, N=30000) # Using Crux.jl
policy_dqn = solve(solver_dqn, mdp) # Error comes here
Stacktrace:
policy_dqn
MethodError: no method matching logpdf(::Distributions.Categorical{Float64, Vector{Float64}}, ::Nothing)
Closest candidates are:
logpdf(::Distributions.DiscreteNonParametric, !Matched::Real) at C:\Users\name\.julia\packages\Distributions\Xrm9e\src\univariate\discrete\discretenonparametric.jl:106
logpdf(::Distributions.UnivariateDistribution{S} where S<:Distributions.ValueSupport, !Matched::AbstractArray) at deprecated.jl:70
logpdf(!Matched::POMDPPolicies.PlaybackPolicy, ::Any) at C:\Users\name\.julia\packages\POMDPPolicies\wMOK3\src\playback.jl:34
...
logpdf(::Crux.ObjectCategorical, ::Float32)#utils.jl:16
logpdf(::Crux.DistributionPolicy, ::Vector{Float64}, ::Float32)#policies.jl:305
var"#exploration#133"(::Base.Iterators.Pairs{Union{}, Union{}, Tuple{}, NamedTuple{(), Tuple{}}}, ::typeof(Crux.exploration), ::Crux.DistributionPolicy, ::Vector{Float64})#policies.jl:302
exploration#policies.jl:297[inlined]
action(::Crux.DistributionPolicy, ::Vector{Float64})#policies.jl:294
var"#exploration#136"(::Crux.DiscreteNetwork, ::Int64, ::typeof(Crux.exploration), ::Crux.MixedPolicy, ::Vector{Float64})#policies.jl:326
var"#step!#173"(::Bool, ::Int64, ::typeof(Crux.step!), ::Dict{Symbol, Array}, ::Int64, ::Crux.Sampler{Main.workspace#2.SliderMDP, Vector{Float32}, Crux.DiscreteNetwork, Crux.ContinuousSpace{Tuple{Int64}}, Crux.DiscreteSpace})#sampler.jl:55
var"#steps!#174"(::Int64, ::Bool, ::Int64, ::Bool, ::Bool, ::Bool, ::typeof(Crux.steps!), ::Crux.Sampler{Main.workspace#2.SliderMDP, Vector{Float32}, Crux.DiscreteNetwork, Crux.ContinuousSpace{Tuple{Int64}}, Crux.DiscreteSpace})#sampler.jl:108
var"#fillto!#177"(::Int64, ::Bool, ::typeof(Crux.fillto!), ::Crux.ExperienceBuffer{Array}, ::Crux.Sampler{Main.workspace#2.SliderMDP, Vector{Float32}, Crux.DiscreteNetwork, Crux.ContinuousSpace{Tuple{Int64}}, Crux.DiscreteSpace}, ::Int64)#sampler.jl:156
solve(::Crux.OffPolicySolver, ::Main.workspace#2.SliderMDP)#off_policy.jl:86
top-level scope#Local: 1[inlined]
Short answer:
Change your output vector to Float32 i.e. Float32[-.1, 0, .1].
Long answer:
Crux creates a Distribution over your network's output values, and at some point (policies.jl:298) samples a random value from it. It then converts this value to a Float32. Later (utils.jl:15) it does a findfirst to find the index of this value in the original output array (stored as objs within the distribution), but because the original array is still Float64, this fails and returns a nothing. Hence the error.
I believe this (converting the sampled value but not the objs array and/or not using approximate equality check i.e. findfirst(isapprox(x), d.objs)) to be a bug in the package, and would encourage you to raise this as an issue on Github.
I have the following when error when trying to use the preProcess function from the caret package. The predict function works for knn and median imputation, but gives an error for bagging. How should I edit my call to the predict function.
Reproducible example:
data = iris
set.seed(1)
data = as.data.frame(lapply(data, function(cc) cc[ sample(c(TRUE, NA), prob = c(0.8, 0.2), size = length(cc), replace = TRUE) ]))
preprocess_values = preProcess(data, method = c("bagImpute"), verbose = TRUE)
data_new = predict(preprocess_values, data)
This gives the following error:
> data_new = predict(preprocess_values, data)
Error in UseMethod("predict") :
no applicable method for 'predict' applied to an object of class "NULL"
The preprocessing/imputation functions in caret work only for numerical variables.
As stated in the help of preProcess
x a matrix or data frame. Non-numeric predictors are allowed but will be ignored.
You most likely found a bug in the part that should ignore the non numerical variables which throws an uninformative error instead of ignoring them.
If you remove the factor variable the imputation works
library(caret)
df <- iris
set.seed(1)
df <- as.data.frame(lapply(data, function(cc) cc[ sample(c(TRUE, NA), prob = c(0.8, 0.2), size = length(cc), replace = TRUE) ]))
df <- df[,-5] #remove factor variable
preprocess_values <- preProcess(df, method = c("bagImpute"), verbose = TRUE)
data_new <- predict(preprocess_values, df)
The last line of code works but results in a bunch of warnings:
In cprob[tindx] + pred :
longer object length is not a multiple of shorter object length
These warnings are not from caret but from the internal call to ipred::bagging which is called internally by caret::preProcess. The cause for these errors are instances in the data where there are 3 NA values in a row, when they are removed
df <- df[rowSums(sapply(df, is.na)) != 3,]
preprocess_values <- preProcess(df, method = c("bagImpute"), verbose = TRUE)
data_new <- predict(preprocess_values, df)
the warnings disappear.
You should check out recipes, and specifically step_bagimpute, to overcome the above mentioned limitations.
What I am trying to do:
I want to connect any Layers from different models to create a new keras model.
What I found so far:
https://github.com/keras-team/keras/issues/4205: using the Model's call class to change the input of another model. My problems with this approach:
Can only change the input of the Model, no other layers. So if I want to cut off some layers at the beginning of the encoder, that is not possible
Not a fan of the nested array structure when getting the config file. Would prefer to have a 1D-array
When using model.summary() or plot_model(), the encoder only shows as "Model". If anything I would say both models should be wrapped. So the config should show [model_base, model_encoder] and not [base_input, base_conv2D, ..., encoder_model]
To be fair, with this approach: https://github.com/keras-team/keras/issues/3021, the point above is actually possible, but again, it is very inflexible. As soon as I want to cut off some layers at the top or bottom of the base or encoder network, this approach fails
https://github.com/keras-team/keras/issues/3465: Adding new layers to a base model by using any output of the base model. Problems here:
While it is possible to use any layer from the base model, which means I can cut off layers from the base model, I can not load the encoder as a keras model. The top models always must be created new.
What I have tried:
My approach to connecting any layers from different models:
Clear inbound nodes of input layer
use the call() method of the output layer with the tensor of the output layer
Clean up the outbound nodes of the output tensor by switching out the new created tensor with the previous output tensor
I was really optimistic at first, as the summary() and the plot_model() got me exactly what I wanted, thus the Node graph should be fine, right? But I ran into errors when training. While the approach in the "What I found so far" section trained fine, I ran into an error with my approach. This is the error message:
File "C:\Anaconda\envs\dlpipe\lib\site-packages\tensorflow\python\framework\op_def_library.py", line 508, in apply_op
(input_name, err))
ValueError: Tried to convert 'x' to a tensor and failed. Error: None values not supported.
Might be an important info, that I am using Tensorflow as backend. I was able to trace back the root of this error. It seems like there is an error when the gradients are calculated. Usually, there is a gradient calculation for each node, but all the nodes of the base network have "None" when using my approach. So basically in keras/optimizers.py, get_updates() when the gradients are calculated (grad = self.get_gradients(loss, params)).
Here is the code (without the training), with all three approaches implemented:
def create_base():
in_layer = Input(shape=(32, 32, 3), name="base_input")
x = Conv2D(32, (3, 3), padding='same', activation="relu", name="base_conv2d_1")(in_layer)
x = Conv2D(32, (3, 3), padding='same', activation="relu", name="base_conv2d_2")(x)
x = MaxPooling2D(pool_size=(2, 2), name="base_maxpooling_2d_1")(x)
x = Dropout(0.25, name="base_dropout")(x)
x = Conv2D(64, (3, 3), padding='same', activation="relu", name="base_conv2d_3")(x)
x = Conv2D(64, (3, 3), padding='same', activation="relu", name="base_conv2d_4")(x)
x = MaxPooling2D(pool_size=(2, 2), name="base_maxpooling2d_2")(x)
x = Dropout(0.25, name="base_dropout_2")(x)
return Model(inputs=in_layer, outputs=x, name="base_model")
def create_encoder():
in_layer = Input(shape=(8, 8, 64))
x = Flatten(name="encoder_flatten")(in_layer)
x = Dense(512, activation="relu", name="encoder_dense_1")(x)
x = Dropout(0.5, name="encoder_dropout_2")(x)
x = Dense(10, activation="softmax", name="encoder_dense_2")(x)
return Model(inputs=in_layer, outputs=x, name="encoder_model")
def extend_base(input_model):
x = Flatten(name="custom_flatten")(input_model.output)
x = Dense(512, activation="relu", name="custom_dense_1")(x)
x = Dropout(0.5, name="custom_dropout_2")(x)
x = Dense(10, activation="softmax", name="custom_dense_2")(x)
return Model(inputs=input_model.input, outputs=x, name="custom_edit")
def connect_layers(from_tensor, to_layer, clear_inbound_nodes=True):
try:
tmp_output = to_layer.output
except AttributeError:
raise ValueError("Connecting to shared layers is not supported!")
if clear_inbound_nodes:
to_layer.inbound_nodes = []
else:
tensor_list = to_layer.inbound_nodes[0].input_tensors
tensor_list.append(from_tensor)
from_tensor = tensor_list
to_layer.inbound_nodes = []
new_output = to_layer(from_tensor)
for out_node in to_layer.outbound_nodes:
for i, in_tensor in enumerate(out_node.input_tensors):
if in_tensor == tmp_output:
out_node.input_tensors[i] = new_output
if __name__ == "__main__":
base = create_base()
encoder = create_encoder()
#new_model_1 = Model(inputs=base.input, outputs=encoder(base.output))
#plot_model(new_model_1, to_file="plots/new_model_1.png")
new_model_2 = extend_base(base)
plot_model(new_model_2, to_file="plots/new_model_2.png")
print(new_model_2.summary())
base_layer = base.get_layer("base_dropout_2")
top_layer = encoder.get_layer("encoder_flatten")
connect_layers(base_layer.output, top_layer)
new_model_3 = Model(inputs=base.input, outputs=encoder.output)
plot_model(new_model_3, to_file="plots/new_model_3.png")
print(new_model_3.summary())
I know this is a lot of text and a lot of code. But I feel like it is needed to explain the issue here.
EDIT: I just tried thenao and I think the error gives away more information:
theano.gradient.DisconnectedInputError:
Backtrace when that variable is created:
It seems like every layer from the encoder model has some connection with the encoder input layer via TensorVariables.
So this is what I ended up with for the connect_layer() function:
def connect_layers(from_tensor, to_layer, old_tensor=None):
# if there is any shared layer after the to_layer, it is not supported
try:
tmp_output = to_layer.output
except AttributeError:
raise ValueError("Connecting to shared layers is not supported!")
# check if to_layer has multiple input_tensors, and therefore some sort of merge layer
if len(to_layer.inbound_nodes[0].input_tensors) > 1:
tensor_list = to_layer.inbound_nodes[0].input_tensors
found_tensor = False
for i, tensor in enumerate(tensor_list):
# exchange the old tensor with the new created tensor
if tensor == old_tensor:
tensor_list[i] = from_tensor
found_tensor = True
break
if not found_tensor:
tensor_list.append(from_tensor)
from_tensor = tensor_list
to_layer.inbound_nodes = []
else:
to_layer.inbound_nodes = []
new_output = to_layer(from_tensor)
tmp_out_nodes = to_layer.outbound_nodes[:]
to_layer.outbound_nodes = []
# recursively connect all layers after the current to_layer
for out_node in tmp_out_nodes:
l = out_node.outbound_layer
print("Connecting: " + str(to_layer) + " ----> " + str(l))
connect_layers(new_output, l, tmp_output)
As each Tensor has all the information about it's root tensor via -> owner.inputs -> owner.inputs -> ..., all tensor following the new_output tensor must be updated.
It was a lot easier to debug that with theano then with tensorflow backend.
I still need to figure out how to deal with shared layers. With the current implementation it is not possible to connect other models that contain a shared layer after the first to_layer.
The following code I've written, fails at self.optimizer.compute_gradients(self.output,all_variables)
import tensorflow as tf
import tensorlayer as tl
from tensorflow.python.framework import ops
import numpy as np
class Network1():
def __init__(self):
ops.reset_default_graph()
tl.layers.clear_layers_name()
self.sess = tf.Session()
self.optimizer = tf.train.AdamOptimizer(learning_rate=0.1)
self.input_x = tf.placeholder(tf.float32, shape=[None, 784],name="input")
input_layer = tl.layers.InputLayer(self.input_x)
relu1 = tl.layers.DenseLayer(input_layer, n_units=800, act = tf.nn.relu, name="relu1")
relu2 = tl.layers.DenseLayer(relu1, n_units=500, act = tf.nn.relu, name="relu2")
self.output = relu2.all_layers[-1]
all_variables = relu2.all_layers
self.gradient = self.optimizer.compute_gradients(self.output,all_variables)
init_op = tf.initialize_all_variables()
self.sess.run(init_op)
with warning,
TypeError: Argument is not a tf.Variable: Tensor("relu1/Relu:0",
shape=(?, 800), dtype=float32)
However when I change that line to tf.gradients(self.output,all_variables), the code works fine, at least no warning is reported. Where did I miss, since I think these two methods are actually executing the same thing, that is return a list of (gradient, variable) pairs.
optimizer.compute_gradients wraps tf.gradients(), as you can see here. It does additional asserts (which explains your error).
I would like to add to the above answer by referring to a simple point. optimizer.compute_gradients return a list of tuples as (grads, vars) pairs. Variables are always there, but the gradients might be None. That makes sense since computing the gradients of specific loss with respect to some of the variables in var_list can be None. It says there is no dependency.
On the other hand, tf.gradients only return the list of sum(dy/dx) for each variable. It MUST be accompanied by the variable list for applying the gradient update.
Henceforth, the following two approaches can be utilized interchangeably:
### Approach 1 ###
variable_list = desired_list_of_variables
gradients = optimizer.compute_gradients(loss,var_list=variable_list)
optimizer.apply_gradients(gradients)
# ### Approach 2 ###
variable_list = desired_list_of_variables
gradients = tf.gradients(loss, var_list=variable_list)
optimizer.apply_gradients(zip(gradients, variable_list))
I am using a kNN to do some classification of labeled images. After my classification is done, I am outputting a confusion matrix. I noticed that one label, bottle was being applied incorrectly more often.
I removed the label and tested again, but then noticed that another label, shoe was being applied incorrectly, but was fine last time.
There should be no normalization, so I'm unsure what is causing this behavior. Testing showed it continued no matter how many labels I removed.
Not totally sure how much code to post, so I'll put some things that should be relevant and pastebin the rest.
def confusionMatrix(classifier, train_DS_X, train_DS_y, test_DS_X, test_DS_y):
# Will output a confusion matrix graph for the predicion
y_pred = classifier.fit(train_DS_X, train_DS_y).predict(test_DS_X)
labels = set(set(train_DS_y) | set(test_DS_y))
def plot_confusion_matrix(cm, title='Confusion matrix', cmap=plt.cm.Blues):
plt.imshow(cm, interpolation='nearest', cmap=cmap)
plt.title(title)
plt.colorbar()
tick_marks = np.arange(len(labels))
plt.xticks(tick_marks, labels, rotation=45)
plt.yticks(tick_marks, labels)
plt.tight_layout()
plt.ylabel('True label')
plt.xlabel('Predicted label')
# Compute confusion matrix
cm = confusion_matrix(test_DS_y , y_pred)
np.set_printoptions(precision=2)
print('Confusion matrix, without normalization')
#print(cm)
plt.figure()
plot_confusion_matrix(cm)
# Normalize the confusion matrix by row (i.e by the number of samples
# in each class)
cm_normalized = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]
print('Normalized confusion matrix')
#print(cm_normalized)
plt.figure()
plot_confusion_matrix(cm_normalized, title='Normalized confusion matrix')
plt.show()
Relevant Code from Main Function:
# Select training and test data
PCA = decomposition.PCA(n_components=.95)
zscorer = ZScoreMapper(param_est=('targets', ['rest']), auto_train=False)
DS = getVoxels (1, .5)
train_DS = DS[0]
test_DS = DS[1]
# Apply PCA and ZScoring
train_DS = processVoxels(train_DS, True, zscorer, PCA)
test_DS = processVoxels(test_DS, False, zscorer, PCA)
print 3*"\n"
# Select the desired features
# If selecting samples or PCA, that must be the only feature
featuresOfInterest = ['pca']
trainDSFeat = selectFeatures(train_DS, featuresOfInterest)
testDSFeat = selectFeatures(test_DS, featuresOfInterest)
train_DS_X = trainDSFeat[0]
train_DS_y = trainDSFeat[1]
test_DS_X = testDSFeat[0]
test_DS_y = testDSFeat[1]
# Optimization of neighbors
# Naively searches for local max starting at numNeighbors
lastScore = 0
lastNeightbors = 1
score = .0000001
numNeighbors = 5
while score > lastScore:
lastScore = score
lastNeighbors = numNeighbors
numNeighbors += 1
#Classification
neigh = neighbors.KNeighborsClassifier(n_neighbors=numNeighbors, weights='distance')
neigh.fit(train_DS_X, train_DS_y)
#Testing
score = neigh.score(test_DS_X,test_DS_y )
# Confusion Matrix Output
neigh = neighbors.KNeighborsClassifier(n_neighbors=lastNeighbors, weights='distance')
confusionMatrix(neigh, train_DS_X, train_DS_y, test_DS_X, test_DS_y)
Pastebin: http://pastebin.com/U7yTs3vs
The issue was in part the result of my axis being mislabeled, when I thought I was removing the faulty label I was in actuality just removing a random label, meaning the faulty data was still being analyzed. Fixing the axis and removing the faulty label which was actually rest yielded:
The code I changed is:
cm = confusion_matrix(test_DS_y , y_pred, labels)
Basically I manually set the ordering based on my list of ordered labels.