I am a Noob. I am currently doing machine learning through a deep neural network, and the input shape and output shape are [36,] and [4,] respectively.

However, there are some problems.
I want to know the input value as the output value in the learning model.
Is there a way or algorithm to go in the reverse direction? Or, in supervised learning, is there an algorithm that can be considered when the size of the output is greater than the input? Please recommend.


thank you..

help

Tried to use GAN model
Tried to use DNN model (multiple activation functions..)
Tried to use Autoencoder model

CodePudding user response：

One way to find large output data from small input in machine learning is to use a model with a high capacity, such as a deep neural network. These models have many layers and a large number of parameters, which allow them to learn complex and non-linear relationships between the input and output data. Another way is to use a generative model such as GANs or VAEs. These models are trained to generate new data that is similar to the training data, and they can be used to generate large amounts of data from a small input. Additionally, one can use data augmentation techniques to generate new data from the existing data set.

Another way is to use a technique called transfer learning, where a pre-trained model is fine-tuned to the specific task at hand. The pre-trained model has already learned to extract useful features from the data, so fine-tuning it on a new task requires less data and can achieve good performance.

It's also important to note that the size of the input data can vary depending on the problem and the type of data, so it's important to ensure that you have enough data to train a model that can generalize well to new examples.

CodePudding user response：

There are a few ways to find big output data from small input in machine learning using Python and deep neural networks (DNNs).

One approach is to use a technique called "data augmentation," which involves creating new training examples by applying various transformations to the original input data. This can be done using Python libraries such as Keras and TensorFlow.

Another approach is to use a technique called "transfer learning," which involves using a pre-trained model to extract features from the input data, and then training a new model on top of those features. This can be done using a variety of pre-trained models, such as those available in the TensorFlow Hub library.

You can also try using Generative Adversarial Networks (GANs), which is a class of deep neural networks that are trained to generate new data that is similar to the input data. This approach can be used to generate new data examples that can be used to train your model.

You can also try using autoencoders, which are neural networks that are trained to reconstruct the input data from a lower-dimensional representation. This can be a powerful technique for finding features in the input data that are relevant to the output.

Finally, you can use dimensionality reduction technique like PCA, t-SNE, UMAP etc on the input data to find the most important features that contribute to the output.

It's important to note that the best approach will depend on the specific task and the nature of the input data. You may need to experiment with different techniques to find the one that works best for your problem.