Questions tagged [neural-networks]
Questions involving defining, training, executing, importing and exporting neural networks.
497
questions
2
votes
0answers
22 views
Does NetChain have a “Distribution” property like PredictorFunction?
PredictorFunction has an incredibly useful "Distribution" property. In the past, this has helped me filter out bad predictions, like this:
...
0
votes
1answer
37 views
Upsampling layer for convolutional autoencoder
I've been trying to build a convolutional autoencoder with Mathematica 12.3. For this, I need to reverse the action of PoolingLayer[]. Is there any layer similar to ...
2
votes
0answers
69 views
Simulated Cartpole Example in Mathematica
I am interested in modifying the code underlying the reinforcement learning example SimulatedCartpole in Mathematica 12.
This example is still listed as Experimental. The help function indicates that ...
1
vote
1answer
76 views
GAN LSTM Time Series
Does anyone know if it is possible to use LSTM or another RNN in GAN architecture as the generator? Here a reference: 1
Thx!
To clarify, what I am asking is it possible in MMA 12.3 (my current version)...
1
vote
0answers
43 views
Inputing pairs of objects to neural networks (eg. pairs of images and sounds)
I'm trying to make a net graph where the input (a pair of objects of distinct but fixed types) will be passed simultaneously trough two net encoders. One will get an array from the image (the first ...
2
votes
1answer
90 views
Neural network evaluate on multiple gpus simultaneously
For instance, I have 4 gpus, now I want to calculate neural network on each gpu simultaneously, or make use of all of my gpus to calculate the data.
The code might like
...
0
votes
0answers
22 views
`FunctionLayer` will calculate a distance matrix
I would like, that a FunctionLayer will calculate a distance matrix between two vectors.
I try a following approach:
...
1
vote
0answers
31 views
Is Parallelization of NetTrain possible?
I was planning to build a Neural Network model and create a list of the same model with the elements of the list being different Initializations of the same model.
Is it possible to Train the entire ...
0
votes
1answer
46 views
Periodic PaddingLayer
The possible values to specify the Padding option of a PaddingLayer include a fixed value, ...
1
vote
0answers
32 views
Should this Dataset be Transposed for best Neural-Net Performance?
I've been working with this Kaggle dataset and intend to use it to make some time-series predictions from it. I've preprocessed it to the point where now I need to start making some decisions about ...
1
vote
0answers
94 views
Is anyone aware of Mathematica implementation of OpenAI's GPT-3 API?
Is anyone aware of Mathematica's implementation of OpenAI's GPT-3 API?
0
votes
0answers
25 views
Decoder for Classifier Network giving error when built using NetGraph
I was trying to design a classifier network with a vector of 4 elements as the input, and a vector of 20 as the output, in the following way:
...
2
votes
1answer
75 views
How to partition variable length vectors via mathematica's neural network?
I need to partition variable length vectors in neural network recently. For instance, the network input is {1,2,3,4,...,2n}, I want to partition two elements as a ...
1
vote
2answers
76 views
How to assemble layers to perform different linear transformations on each column of matrix?
I have a matrix $A_{m×n}$ and I want to assemble a network that performs $n$ different trainable linear transformations $T_{p×m}(i), i\in {1,2,...n}$ on its corresponding columns, (i.e. $B^i = T(i)A^i)...
5
votes
0answers
139 views
How to train custom named entity recognition?
Version 12 has basic NER support for some entities, but how does one recognize a custom entity?
For example, I want to parse text describing products, and parse out three entities: prices, size, and ...
7
votes
1answer
164 views
Running Mathematica in 'SpouseMode'
In this previous question I looked into a deprecated capability of Mathematica from Version 2.
Mathematica command that allows it to read my intentions
Thanks to all for your answers. I note that ...
0
votes
1answer
44 views
How can I rescale data by a learnable constant in a neural network?
I'm working on a class of machine learning problems which fit functions from scalars to vectors. Let's suppose the output of this function is just a set of two-component vectors. I have found that my ...
2
votes
1answer
45 views
NetChain with LinearLayer before a NetFoldOperator, how do I make the memory work?
I have problems using a recurring network built using the NetFoldOperator in cascade to a LinearLayer. What happens is that the ...
3
votes
0answers
41 views
Training a neural network with post-processed data [closed]
I want to fit a large number of functions over a vector space with a neural network. I have a particular architecture in mind, which is naturally broken into two parts: the first part consists of ...
1
vote
1answer
52 views
Trying to train a NetGraph network generates an error
I'm trying to learn neural networks in the Wolfram language, so I've written the following code.
The input is a $2*3$ matrix, split into two parts. The parts are given to distinct linear layers, then ...
6
votes
0answers
111 views
9
votes
1answer
113 views
How to add an extra output port to an existing NetGraph?
I am using a pre-trained neural network which I downloaded from an ONNX repository. I would like to study the output of a given internal layer in the network.
Here is a simple dummy example of what I ...
1
vote
0answers
71 views
Replicating off-the-shelf ODE solvers
I've been working through some literature on ODEs and keep coming across the below pseudo-code expression and am having trouble interpreting it for use in the Wolfram Language. I feel like I've solved ...
5
votes
1answer
144 views
Loss function with regularization
The ML framework has a limited set of loss functions available. How can I extend this by creating my own custom ones?
For example, for L2-regularized training, Mathematica recommends
...
3
votes
0answers
98 views
Classify/NetTrain run indefinitely without output when using GPU
I am experimenting with Mathematica's Classify/NetTrain functions for the first time.
When I don't set a TargetDevice option, Mathematica defaults to the CPU and everything works as expected.
When I ...
1
vote
0answers
37 views
Not enough information was provided to fully specify the output of the NetEncoder
I using a UNET combined with a VGG16 network based loss function. The input of the UNET has an image netencoder attached to it, the output has an image netdecoder attached to it. The loss function ...
4
votes
1answer
260 views
Complex numbers in neural networks
I am trying to construct a neural network which fits a collection of functions from (complex-valued) vectors to complex numbers. After constructing a suitable network, I was surprised to see that ...
5
votes
1answer
77 views
Extracting array product from neural network
I'm training a neural network which is meant to model a collection of functions that map $N$-dimensional vectors to scalars. At this point, I have built up several layers which manipulate the input ...
2
votes
2answers
177 views
Does NetTrain Support Nvidia 30xx series GPU (RTX 3080, 3090 etc)?
Recently, I want to buy some 30xx series GPUs which are used to do deep learning jobs, so does anyone know whether Mathematica's NetTrain supports 30xx series GPUs?
6
votes
1answer
187 views
How to implement this custom NN layer (implement the Accumulate function in the layer)?
I will use this function layer to definte a custom LossFunction in the net.
How do I construct a NetGraph to implement this function?
5
votes
0answers
62 views
Fail to Export/Import seq2seq model (NER) to MXNet or ONNX
Mathematica 12.2 on Linux (Debian)
Here's a simple Named Entity Recognition model:
...
2
votes
1answer
92 views
How to build an RNN with input as character but output a sequence of tokens?
Let’s assume I have “abbcab”-> {“Apple”,”Banana”}, “Bibb”-> {“Carrot”} so the output is a sequence of various length, but I can pad it to be the same length. My problem, how do I create a ...
2
votes
1answer
31 views
What's the difference between train and test evaluation modes of a BatchNormalizationLayer?
After creating a BatchNormalizationLayer:
batchnorm = NetInitialize@BatchNormalizationLayer["Input" -> 3]
We can evaluate it in "Test" or &...
1
vote
1answer
68 views
How is the loss computed from the batch and round?
This is cross-posted from the forum.
I am trying to just reproduce the loss computed during training a very simple network. In this case the loss is just the standard L2 loss function.
Start with some ...
2
votes
1answer
66 views
Copying Weights using NetTrain in Mathematica
I would like to implement the PPO algorithm - Reinforcement Learning - in Mathematica. And for that, I need to be able to copy network weights from say subnetwork NN1 to subnetwork NN2 as an update ...
3
votes
0answers
36 views
Can NetNestOperator take a non pre-defined number of iterations?
I'd like to Nest a particular NetChain inside a larger NetGraph for a random number of ...
2
votes
0answers
27 views
Classify command option question
I'm using Classify to predict who is going to win a battle based on their strength. The input is something like {{"Name 1", Strength1},{"Name 2", Strength2},{"Name 3", ...
2
votes
0answers
58 views
Questions on NetMapOperator, NetNestOperator, NetFoldOperator [closed]
I've been playing around with the Higher-Order Network Construction functionality such as NetMapOperator, NetNestOperator, and <...
11
votes
0answers
190 views
Tutorial example NetTrain fails miserably with a GPU
I have been experimenting with GPU NetTrain on AWS now when Mma 12.2 supports remote batch jobs (I don't have an Nvidia GPU to try these things out otherwise). In ...
0
votes
1answer
75 views
Constructing Net problem
so I'm trying build a simple net that will help me find the winner in a competition. I would like to use as input "Name", "FA" and "Force", and obtain "Winner" ...
2
votes
2answers
164 views
how to implement training for UNET implementation on Wolfram Neural Net Repository?
My question is with reference to the U-NET implementation present on the Wolfram Neural Net Repository
The construction notebook present on the page (link: http://...
1
vote
0answers
61 views
Export and Import trained neural net
I have trained a neural network and would like to save it to disk. I am using Wolfram version 12.1.1.0 on Windows (64-bit). I have entered the following commands:
...
15
votes
1answer
269 views
Minibatch Standard Deviation Layer
I'm reworking some of the GANs I originally made in TensorFlow2 to see if I can improve performance in Mathematica, and have been stuck on how to create a custom Minibatch Standard Deviation Layer. I'...
4
votes
1answer
135 views
What is specifically the correct Generator output type?
I was constructing a GAN to analyze samples of timeseries with 3 features using a previous working template, and ran into the strangest error I haven't seen before. Some of my research so far suggests ...
4
votes
0answers
75 views
How to train each layer in a Neural Network so they optimize different loss functions in an adversarial network?
Example NetGraph to illustrate the idea (Input is an online signal with a value between -1 and 1, Noise is Gaussian Centered at 0 and standard deviation 0.1, EvilNet is constrained to output a value ...
1
vote
0answers
111 views
How to change/reinstall neural network runtime version?
I'm trying to run Mathematica on Azure instance with Tesla K80.
Setting TargetDevice->"GPU" produces the following set of errors for documentation page ...
2
votes
1answer
76 views
Want neural net example of rules {vector->string}
I am using data samples from a multispectral color sensor for color classification as an exercise in beginning to learn about neural networks.
I have succeeded with ...
2
votes
1answer
116 views
Help with creating the NetChain
I'm trying to train a network to take 3 integers and give an output = Max/Min. For example {10, 20, 30} -> 3. But I cannot get the setup correctly. I must be missing something.
...
8
votes
1answer
207 views
Deconvolution approaches for 3D data
Building neural nets in the WL is great. I was in the process of rewriting a previous GAN of mine in the WL to see if I can get better performance, and was surprised to find that the standard ...
1
vote
1answer
72 views
How to improve translating a list of points into a list of 3D volumes
I have a challenging transform I'm trying to accomplish as I try to improve the performance of my 3D GAN.
Background:
I've been working with data from Berkeley's PEER Ground Motion Database to ...
