Say I have one batch that I want to train my model on. Do I simply run tf.Session()'s sess.run(batch) once, or do I have to iterate through all of the batch's examples with a loop in the session? I'm looking for the optimal way to iterate/update the training ops, such as loss. I thought tensorflow would handle it itself, especially in the cases where tf.nn.dynamic_rnn() takes in a batch dimension for listing the examples. I thought, perhaps naively, that a for loop in the python code would be the inefficient method of updating the loss. I am using tf.losses.mean_squared_error(batch) for a regression problem.
My regression problem is given two lists of word vectors (300d each), and determines the similarity between the two lists on a continuous scale from [0, 5]. My supervised model is Deepmind's Differential Neural Computer (DNC). The problem is I do not believe it is learning anything. this is due to the fact that the all of the output from the model is centered around 0 and even negative. I do not know how it could possibly be negative given no negative labels provided. I only call sess.run(loss) for the single batch, I do not create a python loop to iterate through it.
So, what is the most efficient way to iterate the training of a model and how do people go about it? Do they really use python loops to do multiple calls to sess.run(loss) (this was done in the training file example for DNC, and I have seen it in other examples as well). I am certain I get the final loss from the below process, but I am uncertain if the model has actually been trained entirely just because the loss was processed in one go. I also do not understand the point of update_ops returned by some functions, and am uncertain if they are necessary to ensure the model has been trained.
Example of what I mean by processing a batch's loss once:
# assume the model has been defined prior through batch_output_logits
train_loss = tf.losses.mean_squared_error(labels=target,
predictions=batch_output_logits)
with tf.Session() as sess:
sess.run(init_op) # pseudo code, unnecessary for question
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)
# is this the entire batch's loss && model has been trained for that batch?
loss_np = sess.run(train_step, train_loss)
coord.request_stop()
coord.join(threads)
Any input on why I am receiving negative values when the labels are in the range [0, 5] is welcomed as well(general abstract answers for this are fine, because its not the main focus). I am thinking of attempting to create a piece-wise function, if possible, for my loss, so that for any values out of bounds face a rapidly growing exponential loss function. Uncertain how to implement, or if it would even work.
Code is currently private. Once allowed, I will make the repo public.
To run DNC model, go to the project/ directory and run python -m src.main. If there are errors you encounter feel free to let me know.
This model depends upon Tensorflow r1.2, most recent Sonnet, and NLTK's punkt for Tokenizing sentences in sts_handler.py and tests/*.
In a regression model, the network calculates the model output based on the randomly initialized values for your model parameters. That's why you're seeing negative values here; you haven't trained your model enough for it to learn that your values are only between 0 and 5.
Unless I'm missing something, you are only calculating the loss, but you aren't actually training the model. You should probably be calling sess.run(optimizer) on an optimizer, not on your loss function.
You probably need to train your model for multiple epochs (training your model for one epoch = training your model once on the entire dataset).
Batches are used because it is more computationally efficient to train your model on a batch than it is to train it on a single example. However, your data seems to be small enough that you won't have that problem. As such, I would recommend reducing your batch size to as low as possible. As a general rule, you get better training from a smaller batch size, at the cost of added computation.
If you post all of your code, I can take a look.
Related
I am trying to make a CNN model on IAM handwritten words data(which has images of words handwritten by multiple people and targets are text in the images). So, I can encode words to numbers(A=0, B=1 and so on for capital, small and punctuation). But, I couldn't quite decide on what loss to use for this problem, in tensforflow.keras.
I have found out about something called CTC loss on some repositories on GitHub. But, ctc loss was used with softmax as activation in final dense layer. And since softmax outputs probabilities for each class that sum up to 1; So, I presume softmax can't output probabilties for two "o"s at the same time.
So, what are the proper activation and loss functions that I can use for a problem like this one?
I am creating a simple sequential Keras model which will take 10k inputs in a batch of 100. Each input has 3 columns and the corresponding output is sum of that row.
Sequential model has 2 layers- LSTM(Stateful=true) , Dense.
Now, after compiling and fitting the model, I am saving it in 'model.h5' file.
Then, I read the saved model, and call model.predict with a test data (size=10k , batch_size = 100).
Problem: the prediction doesn't work properly for first 400-500 inputs and for the rest its working perfectly fine with very low val_loss.
Case1: I make the LSTM layer Stateless(i.e. Stateful=False)
In this case Keras is providing very accurate outputs for all the test data.
Case2: Instead of saving and then reading again, if I directly apply model.predict on the model created, all the outputs are coming accurately.
But, I need Stateful=True, also, I want to save my model and then resume work on that model later.
1.Is there any way to solve this?
2.Also, when I am providing test data, how is the model's accuracy increasing? ( because the first 400-500 tests provide inaccurate results and the rest are pretty accurate)
Your problem seems to come from losing the hidden states of your cells. During model building they might be reset and this might cause the problem.
So (it's a little bit cumbersome), but you could save and load also a states of your network:
How to save? (assuming that i-th layer is a recurrentone):
hidden_state = model.layers[i].states[0].eval()
cell_state = model.layers[i].states[0].eval()
numpy.save("some name", hidden_state)
numpy.save("some other name", cell_state)
Now when you can reload the hidden state, here you can read on how to set the hidden state in a layer.
Of course - it's the best to pack all of this methods in some kind of object and e.g. class constructor methods.
I am trying to predict whether a particular service ticket raised by client needs a code change.
I have training data.
I have around 17k data points with problem description and tag (Y for code change required and N for no code change)
I did TF-IDF and it gave me 27k features. So I tried to fit RandomForestClassifier (sklearn python) with this 17k x 27k matrix.
I am getting very low scores on test set while training accuracy is very high.
Precision on train set: 89%
Precision on test set: 21%
Can someone suggest any workarounds?
I am using this model now:
sklearn.RandomForestClassifier(n_jobs=3,n_estimators=100,class_weight='balanced',max_features=None,oob_score=True)
Please help!
EDIT:
I have 11k training data with 900 positives (skewed). I tried LinearSVC sparsify but didn't work as well as Truncated SVD (Latent Semantic Indexing). maxFeatures=None performs better on the test set than without it.
I have also tried SVM, logistic (l2 and l1), ExtraTrees. RandonForest still is working best.
Right now, going at 92% precision on positives but recall is 3% only
Any other suggestions would be appreciated!
Update:
Feature engineering helped a lot. I pulled features out of the air (len of chars, len of words, their, difference, ratio, day of week the problem was of reported, day of month, etc) and now I am at 19-20% recall with >95% accuracy.
Food for your thoughts on using word2vec average vectors as deep features for the free text instead of tf-idf or bag of words ???
[edited]
Random forest handles more features than data points quite fine. RF is e.g. used for micro-array studies with e.g. a 100:5000 data point/feature ratio or in single-nucleotide_polymorphism(SNP) studies with e.g 5000:500,000 ratio.
I do disagree with the diagnose provided by #ncfirth, but the suggested treatment of variable selection may help anyway.
Your default random forest is not badly overfitted. It is just not meaningful to pay any attention to a non-cross validated training set prediction performance for a RF model, because any sample will end in the terminal nodes/leafs it has itself defined. But the overall ensemble model is still robust.
[edit] If you would change the max_depth or min_samples_split, the training precision would probably drop, but that is not the point. The non-cross validated training error/precision of a random forest model or many other ensemble models simply does not estimate anything useful.
[I did before edit confuse max_features with n_estimators, sry I mostly use R]
Setting max_features="none" is not random forest, but rather 'bagged trees'. You may benefit from a somewhat lower max_features which improve regularization and speed, maybe not. I would try lowering max_features to somewhere between 27000/3 and sqrt(27000), the typical optimal range.
You may achieve better test set prediction performance by feature selection. You can run one RF model, keep the top ~5-50% most important features and then re-run the model with fewer features. "L1 lasso" variable selection as ncfirth suggests may also be a viable solution.
Your metric of prediction performance, precision, may not be optimal in case unbalanced data or if the cost of false-negative and false-positive is quite different.
If your test set is still predicted much worse than the out-of-bag cross-validated training set, you may have problems with your I.I.D. assumptions that any supervised ML model rely on or you may need to wrap the entire data processing in an outer cross-validation loop, to avoid over optimistic estimation of prediction performance due to e.g. the variable selection step.
Seems like you've overfit on your training set. Basically the model has learnt noise on the data rather than the signal. There are a few ways to combat this, but it seems fairly obvious that you're model has overfit because of the incredibly large number of features you're feeding it.
EDIT:
It seems I was perhaps too quick to jump to the conclusion of overfitting, however this may still be the case (left as an exercise to the reader!). However feature selection may still improve the generalisability and reliability of your model.
A good place to start for removing features in scikit-learn would be here. Using sparsity is a fairly common way to perform feature selection:
from sklearn.svm import LinearSVC
from sklearn.feature_selection import SelectFromModel
import numpy as np
# Create some data
X = np.random.random((1800, 2700))
# Boolean labels as the y vector
y = np.random.random(1800)
y = y > 0.5
y = y.astype(bool)
lsvc = LinearSVC(C=0.05, penalty="l1", dual=False).fit(X, y)
model = SelectFromModel(lsvc, prefit=True)
X_new = model.transform(X)
print X_new.shape
Which returns a new matrix of shape (1800, 640). You can tune the number of features selected by altering the C parameter (called the penalty parameter in scikit-learn but sometimes called the sparsity parameter).
i'm using weka to do some text mining, i'm a little bit confused so i'm here to ask how can i ( with a set of comments that are in a some way classified as: notes, status of work, not conformity, warning) predict if a new comment belong to a specific class, with all the comment (9551) i've done a preprocess obtaining with the filter "stringtowordvector" a vector of tokens, and then i've used the simple kmeans to obtain a number of cluster.
So the question is: if a user post a new comment can i predict with those data if it belong to a category of comment?
sorry if my question is a little bit confused but so am i.
thank you
Trivial Training-validation-test
Create two datasets from your labelled instances. One will be training set and the other will be validation set. The training set will contain about 60% of the labelled data and the validation will contain 40% of the labelled data. There is no hard and fast rule for this split, but a 60-40 split is a good choice.
Use K-means (or any other clustering algorithm) on your training data. Develop a model. Record the model's error on training set. If the error is low and acceptable, you are fine. Save the model.
For now, your validation set will be your test dataset. Apply the model you saved on your validation set. Record the error. What is the difference between training error and validation error? If they both are low, the model's generalization is "seemingly" good.
Prepare a test dataset where you have all the features of your training and test dataset but the class/cluster is unknown.
Apply the model on the test data.
10-fold cross validation
Use all of your labelled data instances for this task.
Apply K-means (or any other algorithm of your choice) with a 10-fold CV setup.
Record the training error and CV error. Are they low? Is the difference between the errors is low? If yes, then save the model and apply it on the test data whose class/cluster is unknown.
NB: The training/test/validation errors and their differences will give you an "very initial" idea of overfitting/underfitting of your model. They are sanity tests. You need to perform other tests like learning curves to see if your model overfits or underfits or perfect. If there appears to be an overfitting and underfitting problem, you need to try many different techniques to overcome them.
I'm using Weka and would like to perform regression with random forests. Specifically, I have a dataset:
Feature1,Feature2,...,FeatureN,Class
1.0,X,...,1.4,Good
1.2,Y,...,1.5,Good
1.2,F,...,1.6,Bad
1.1,R,...,1.5,Great
0.9,J,...,1.1,Horrible
0.5,K,...,1.5,Terrific
.
.
.
Rather than learning to predict the most likely class, I want to learn the probability distribution over the classes for a given feature vector. My intuition is that using just the RandomForest model in Weka would not be appropriate, since it would be attempting to minimize its absolute error (maximum likelihood) rather than its squared error (conditional probability distribution). Is that intuition right? Is there a better model to be using if I want to perform regression rather than classification?
Edit: I'm actually thinking now that in fact it may not be a problem. Presumably, classifiers are learning the conditional probability P(Class | Feature1,...,FeatureN) and the resulting classification is just finding the c in Class that maximizes that probability distribution. Therefore, a RandomForest classifier should be able to give me the conditional probability distribution. I just had to think about it some more. If that's wrong, please correct me.
If you want to predict the probabilities for each class explicitly, you need different input data. That is, you would need to replace the value to predict. Instead of one data set with the class label, you would need n data sets (for n different labels) with aggregated data for each unique feature vector. Your data would look something like
Feature1,...,Good
1.0,...,0.5
0.3,...,1.0
and
Feature1,...,Bad
1.0,...,0.8
0.3,...,0.1
and so on. You would need to learn one model for each class and run them separately on any data to be classified. That is, for each label you learn a model to predict a number that is the probability of being in that class, given a feature vector.
If you don't need the probabilities to be predicted explicitly, have a look at the Bayesian classifiers in Weka, which make use of probabilities in the models that they learn.