Image Generator for Tabular Data (IGTD) transforms tabular data into images. The algorithm assigns each feature to a pixel in the image. According to the assignment, an image is generated for each data sample, in which the pixel intensity reflects the value of the corresponding feature in the sample. The algorithm searches for an optimized assignment of features to pixels by minimizing the difference between the ranking of pairwise distances between features and the ranking of pairwise distances between the assigned pixels, where the distances between pixels are calculated based on their coordinates in the image. Minimizing the difference between the two rankings assigns similar features to neighboring pixels and dissimilar features to pixels that are far apart. The optimization is achieved through an iterative process of swapping the pixel assignments of two features. In each iteration, the algorithm identifies the feature that has not been considered for swapping for the longest time, and seeks for a feature swapping for it that reduces the difference between the two rankings most.
For the sbusequent prediction modeling, Convolutional Neural Networks (CNNs) can be trained based on the images generated by IGTD. The provided code allows the input of one or more data modalities. The images of each data modality is input into multiple convolution layers for embedding. Then the embeddings of multiple data modalities are concatenated and forwarded to multiple dense layers to make prediction.
To set up the Python environment needed to run this algorithm:
- Install conda package manager.
- Clone this repository.
- Create the environment as shown below.
conda env create -f environment.yml -n IGTD conda activate IGTD pip install git+https://github.com/ECP-CANDLE/candle_lib@develop