I try to get reliable features for ImageNet to do further classification on them. To achieve that I would like to use tensorflow with Alexnet, for feature extraction. That means I would like to get the values from the last layer in the CNN. Could someone write a piece of Python code that explains how that works?
As jonrsharpe mentioned, that's not really stackoverflow's MO, but in practice, many people do choose to write code to help explain answers (because it's often easier).
So I'm going to assume that this was just miscommunication, and you really intended to ask one of the following two questions:
How does one grab the values of the last layer of Alexnet in TensorFlow?
How does feature extraction from the last layer of a deep convolutional network like alexnet work?
The answer to the first question is actually very easy. I'll use the cifar10 example code in TensorFlow (which is loosely based on AlexNet) as an example. The forward pass of the network is built in the inference function, which returns a variable representing the output of the softmax layer. To actually get predicted image labels, you just argmax the logits, like this: (I've left out some of the setup code, but if you're already running alexnet, you already have that working)
logits = cifar10.inference(images)
predictions = tf.argmax(logits,1)
# Actually run the computation
labels = session.run([predictions])
So grabbing just the last layer features is literally just as easy as asking for them. The only wrinkle is that, in this case, cifar10 doesn't natively expose them, so you need to modify the cifar10.inference function to return both:
# old code in cifar10.inference:
# return softmax_linear
# new code in cifar10.inference:
return softmax_linear, local4
And then modify all the calls to cifar10.inference, like the one we just showed:
logits,local4 = cifar10.inference(images)
predictions = tf.argmax(logits,1)
# Actually run the computation, this time asking for both answers
labels,last_layer = session.run([predictions, local4])
And that's it. last_layer contains the last layer for all of the inputs you gave the model.
As for the second question, that's a much deeper question, but I'm guessing that's why you want to work on it. I'd suggest starting by reading up on some of the papers published in this area. I'm not an expert here, but I do like Bolei Zhou's work. For instance, try looking at Figure 2 in "Learning Deep Features for Discriminative Localization". It's a localization paper, but it's using very similar techniques (and several of Bolei's papers use it).
Related
Hi everybody I have this basic question that isn't related to any particular piece of code
I'm following some papers about the use of CNN to assess the quality of images and, in all of them, the networks that are being used are the ones offered by keras-applications (VGG16-19, ResNet..) with some variations depanding on the paper
Often, in these works, the variations to the basic networks aren't bound to them but valid for the whole paper, meaning the only differences should be the images input size and the specific image pre-processing function, the latters also provided within keras-application."basic network name"
My question is if what I'm doing is sufficient because, in fact, when I try to replicate the results of a paper, there are some networks that underperform (the results aren't that bad, meaning the network is still training in the right direction) compared to the expected results but others don't. In particular, I had problems with ResNet50 and VGG16 but not with MobileNet v1/v2 and InceptionV3: no matter how much learning rate (or even the dropouts) I use, the validation loss obtained at the end of the trainings are almost 10% worse to the published results when using ResNet or VGG
I'm sure the code is correct, as I said the only difference when changing the loaded basic network is which image pre-processing function is selected. At this point I have two possible ideas:
I'm using a different setup compared to the papers (tensorflow 1.15) and the Keras-applications has some kind of bug. In fact the last published version of that module has some bugs regarding the pre-processing in that the code threw an exception when I tried it for the first time. By digging on their Git page I found that that problem was endeed a bug that had already been fixed within the last commit (not yet published inside a proper release). Sadly, that doesn't mean its behaviour isn't bugged but I can't verify that by myself
Some networks, such as VGG16 and ResNet50, have more requirements to work properly and I'm missing them, what can you tell me from your experience? Have you found yourself in a situation like mine? Note that I'm not talking about parameters such as the dropouts, learning rates etc.. because they are provided by the papers, I'm wondering if for example the function, and relative interpolation, used to load the images could matter in any kind of way
I am working on a sequence prediction problem where my inputs are of size (numOfSamples, numOfTimeSteps, features) where each sample is independent, number of time steps is uniform for each sample (after pre-padding the length with 0's using keras.pad_sequences), and my number of features is 2. To summarize my question(s), I am wondering how to structure my Y-label dataset to feed the model and want to gain some insight on how to properly structure my model to output what I want.
My first feature is a categorical variable encoded to a unique int and my second is numerical. I want to be able to predict the next categorical variable as well as an associated feature2 value, and then use this to feed back into the network to predict a sequence until the EOS category is output.
This is a main source I've been referencing to try and understand how to create a generator for use with keras.fit_generator.
[1]
There is no confusion with how the mini-batch for "X" data is grabbed, but for the "Y" data, I am not sure about the proper format for what I am trying to do. Since I am trying to predict a category, I figured a one-hot vector representation of the t+1 timestep would be the proper way to encode the first feature, I guess resulting in a 4? Dimensional numpy matrix?? But I'm kinda lost with how to deal with the second numerical feature.
Now, this leads me to questions concerning architecture and how to structure a model to do what I am wanting. Does the following architecture make sense? I believe there is something missing that I am not understanding.
Proposed architecture (parameters loosely filled in, nothing set yet):
model = Sequential()
model.add(Masking(mask_value=0., input_shape=(timesteps, features)))
model.add(LSTM(hidden_size, return_sequences=True))
model.add(TimeDistributed(Dense(vocab_size)))
model.add(Activation('softmax'))
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['categorical_accuracy'])
model.fit_generator(...) #ill figure this out
So, at the end, a softmax activation can predict the next categorical value for feature1. How do I also output a value for feature2 so that I can feed the new prediction for both features back as the next time-step? Do I need some sort of parallel architecture with two LSTMs that are combined somehow?
This is my first attempt at doing anything with neural networks or Keras, and I would not say I'm "great" at python, I can get by though. However, I feel I have a decent grasp at the fundamental theoretical concepts, but lack the practice.
This question is sorta open ended, with encouragement to pick apart my current strategy.
Once again, the overall goal is to predict both features (categorical, numeric) in order to predict "full sequences" from intermediate length sequences.
Ex. I train on these padded max-len sequences, but in production I want to use this to predict the remaining part of the currently unseen time-steps, which would be variable length.
Okay, so If I understand you properly (correct me if I'm wrong) you would like to predict next features based on the current ones.
When it comes to categorical variables, you are on point, your Dense layer should output N-1 vector containing probability of each class (while we are at it, if you, by any chance, use pandas.get_dummies remember to specify argument drop_first=True, similiar approach should be employed whatever you are using for one-hot encoding).
Except those N-1 output vector for each sample, it should output one more number for numerical value.
Remember to output logits (no activation, don't use softmax at the end like you currently do). Afterwards network output should be separated into N-1 part (your categorical feature) and passed to loss function able to handle logits (e.g. in Tensorflow it is tf.nn.softmax_cross_entropy_with_logits_v2 which applies numerically stable softmax for you).
Now, your N-th element of network output should be passed to different loss, probably Mean Squared Error.
Based on loss value of those two losses (you could take a mean of both to obtain one loss value), you backpropagate through the network and it might do just fine.
Unfortunately I'm not skilled enough in Keras in order to help you with the code, but I think you will figure it out yourself. While we're at it, I would like to suggest PyTorch for more custom neural networks (I think yours fits this description), though it's definitely doable in Keras as well, your choice.
Additional 'maybe helpful' thought: you may check Teacher Forcing for your kind of task. More on the topic and theory behind it can be found in the outstanding Deep Learning Book and code example (though in PyTorch once again), can be found in their docs here.
BTW interesting idea, mind if I use it in connection with my current research trajectory (with kudos going to you of course)? Comment on this answer if so we can talk it out in chat.
Basically every answer I was looking for was exampled and explained in this tutorial. Absolutely great resource for trying to understand how to model multi-output networks. This one goes through a lengthy walkthrough of a multi-output CNN architecture. It only took me about three weeks to stumble upon, however.
https://www.pyimagesearch.com/2018/06/04/keras-multiple-outputs-and-multiple-losses/
I have created a prediction model and used RNN in it offered by the tensorflow library in Python. Here is the complete code I have created and tried:
Jupyter Notbook of the Code
But I have doubts.
1) Whether RNN is correct for what I am trying to predict?
2) Is there a better algorithm I can try?
3) Can anyone suggest me how I can give multiple inputs and get the necessary output using tensorflow model? Can anyone guide me please.
I hope I am clear on my points. Please do tell me if anything else required.
Having doubts is normal, but you should try to measure them before asking for advice. If you don't have a clear thing you want to improve it's unlikely you will get something better.
1) Whether RNN is correct for what I am trying to predict?
Yes. RNN is used appropriately here. If you don't care much about having arbitrary length input sequences, you can also try to force them to a fixed size and then apply convolutions on top (see convolutional NeuralNetworks), or even try with a more simple DNN.
The more important question to ask yourself is if you have the right inputs and if you have sufficient training data to learn what you hope to learn.
2) Is there a better algorithm I can try?
Probably no. As I said RNN seems appropriate for this problem. Do try some hyper parameter tuning to make sure you don't accidentally just pick a sub-optimal configuration.
3) Can anyone suggest me how I can give multiple inputs and get the necessary output using tensorflow model? Can anyone guide me please.
The common way to handle variable length inputs is to set a max length and pad the shorter examples until they reach that length. The max length can be a variable you pick or you can dynamically set it to the largest length in the batch. This is needed only because the internal operations are done in batches. You can pick which results you want. Picking the last one is reasonable (the model will just have to learn to propagate the state for the padding values). Another reasonable thing to do is to pick the first one you get after feeding the last meaningful value into the RNN.
Looking at your code, there's one thing I would improve:
Instead of computing a loss on the last value only, I would compute it over all values in the series. This gives your model more training data with very little performance degradation.
I am trying to follow the tutorial for Language Modeling on the TensorFlow site. I see it runs and the cost goes down and it is working great, but I do not see any way to actually get the predictions from the model. I tried following the instructions at this answer but the tensors returned from session.run are floating point values like 0.017842259, and the dictionary maps words to integers so that does not work.
How can I get the predicted word from a tensorflow model?
Edit: I found this explanation after searching around, I just am not sure what x and y would be in the context of this example. They don't seem to use the same conventions for this example as they do in the explanation.
The tensor you are mentioning is the loss, which defines how the network is training. For prediction, you need to access the tensor probabilities which contain the probabilities for the next word. If this was classification problem, you'd just do argmax to get the top probability. But, to also give lower probability words a chance of being generated,some kind of sampling is often used.
Edit: I assume the code you used is this. In that case, if you look at line 148 (logits) which can be converted into probabilities by simply applying the softmax function to it -- like shown in the pseudocode in tensorflow website. Hope this helps.
So after going through a bunch of other similar posts I figured this out. First, the code explained in the documentation is not the same as the code on the GitHub repository. The current code works by initializing models with data inside instead of passing data to the model as it goes along.
So basically to accomplish what I was trying to do, I reverted my code to commit 9274f5a (also do the same for reader.py). Then I followed the steps taken in this post to get the probabilities tensor in my run_epoch function. Additionally, I followed this answer to pass the vocabulary to my main function. From there, I inverted the dict using vocabulary = {v: k for k, v in vocabulary.items()} and passed it to run_epoch.
Finally, we can get the predicted word in run_epoch by running current_word = vocabulary[np.argmax(prob, 1)] where prob is the tensor returned from session.run()
Edit: Reverting the code as such should not be a permanent solution and I definitely recommend using #Prophecies answer above to get the probabilities tensor. However, if you want to get the word mapping, you will need to pass the vocabulary as I did here.
I'm in need of an artificial neural network library (preferably in python) for one (simple) task. I want to train it so that it can tell wether a thing is in an image. I would train it by feeding it lots of pictures and telling it wether it contains the thing I'm looking for or not:
These images contain this thing, return True (or probability of it containing the thing)
These images do not contain this thing, return False (or probability of it containing the thing)
Does such a library already exist? I'm fairly new to ANNs and image recognition; although I understand how they both work in principle I find it quite hard to find an adequate library for this task, and even research in this field has proven to be kind of a frustration - any advice towards the right direction is greatly appreciated.
There are several good Neural Network approaches in Python, including TensorFlow, Caffe, Lasagne, and sknn (Sci-kit Neural Network). sknn provides an easy, out of the box solution, although in my opinion it is more difficult to customize and can be slow on large datasets.
One thing to consider is whether you want to use a CNN (Convolutional Neural Network) or a standard ANN. With an ANN you will mostly likely have to "unroll" your images into a vector whereas with a CNN, it expects the image to be a cube (if in color, a square otherwise).
Here is a good resource on CNNs in Python.
However, since you aren't really doing a multiclass image classification (for which CNNs are the current gold standard) and doing more of a single object recognition, you may consider a transformed image approach, such as one using the Histogram of Oriented Gradients (HOG).
In any case, the accuracy of a Neural Network approach, especially when using CNNs, is highly dependent on successful hyperparamter tuning. Unfortunately, there isn't yet any kind of general theory on what hyperparameter values (number and size of layers, learning rate, update rule, dropout percentage, batch size, etc.) are optimal in a given situation. So be prepared to have a nice Training, Validation, and Test set setup in order to fit a robust model.
I am unaware of any library which can do this for you. I use a lot of Caffe and can give you a solution till you find a single library which can do it for you.
I hope you know about ImageNet and that Caffe has a trained model based on ImageNet.
Here is the idea:
Define what the object is. Say object = "laptop".
Use Caffe's ImageNet trained model, change the code to display the required output you want (you mentioned TRUE or FALSE) when the object is in the output labels.
Here is a link to the ImageNet tutorial which I wrote.
Here is what you might try:
Take a look here. It is a stripped down version of the ImageNet program which I used in a prediction engine.
In line 80 you'll get the top-1 predicted output label. In line 86 you'll get the top-5 predicted labels. Write a line of code to check whether object is in the output_label and return TRUE or FALSE according to it.
I understand that you are looking for a specific library, I will look for it, but this is something I would try out in the beginning.