# Clustering and Classification Networks

## TL;DR

This paper introduces a novel network architecture that enhances clustering and classification performance through a three-step process involving architecture search, class clustering, and reclassification, achieving state-of-the-art results on datasets like Cifar-100.

## Contribution

The paper proposes a new network design with a three-level fully connected layer division and a recursive clustering approach for improved classification accuracy.

## Key findings

- Achieved 11.56% error rate on Cifar-100.
- Demonstrated state-of-the-art performance with the proposed method.
- Validated effectiveness across various dataset sizes.

## Abstract

In this paper, we will describe a network architecture that demonstrates high performance on various sizes of datasets. To do this, we will perform an architecture search by dividing the fully connected layer into three levels in the existing network architecture. The first step is to learn existing CNN layer and existing fully connected layer for 1 epoch. The second step is clustering similar classes by applying L1 distance to the result of Softmax. The third step is to reclassify using clustering class masks. We accomplished the result of state-of-the-art by performing the above three steps sequentially or recursively. The technology recorded an error of 11.56% on Cifar-100.