Improving Image Classification Accuracy Through Cluster Optimization in Latent Space

In this paper, we propose a novel image classification method that combines a variational encoder and triplet loss. The method aims to improve classification accuracy by minimizing the variance of clusters in the latent space. The variational encoder transforms data into the latent space to form probabilistic distributions, which represent clusters of the categories we want to classify. Triplet loss reduces the overlap between these clusters, leading to accurate boundary setting and ensuring that samples move toward the cluster centers. This research contributes to improving the performance of image classification tasks by combining the strengths of these two approaches. The proposed method was implemented using PyTorch to create a neural network for experimentation, utilizing the simplest linear layer for proof of concept, and was evaluated on the MNIST dataset. The results show that the proposed method outperforms existing methods by achieving an average accuracy of 98.10%, demonstrating a performance gain of more than 0.2% compared to previous methods.