Convolutional Neural Network (CNN) on numerical data

Hello, I am a newbie to deep learning.
I have numerical data with 300 rows and 667 columns. The last column contains labels with the help of which, I want to classify the datasets into two classes. I am having problems with data input to the convolutional layer. I am having difficulty applying filters to numerical data.
Can anyone please suggest a proper implementation of CNN to numerical data to me?

P.S.: I know that the dataset is small. I have data augmentation in the to-do task but before that, I want to understand the concepts related to the input layer.

Wouldn’t a Mutil-Layered Perceptron be more suitable for your task at hand? I’m not sure how a CNN could be implemented properly.

1 Like

Thank you so much for the response.
I have already implemented MLP. I want to go a step ahead and want to know what changes in accuracy and performance can CNN make.

A CNN assumes there is some spatial similarity or continuity between pixels. What is the relationship between the rows and columns in your dataset? A CNN is not just a magical recipe.

Thank you for sharing the information.
Each row is a spectrum of cells. So, each row with the mentioned number of columns can give spectral information of respective cells.