Earticle

Data-driven learning is the most representative deep learning method. Generative adversarial networks (GANs) are designed to generate sufficient data to support such learning. The learning process of GAN models typically trains a generator and discriminator in turn. However, overfitting problems occur when the discriminator depends excessively on the training data. When this problem persists, the image created by the generator shows a similar appearance to the learning image. Images similar to learning images eventually lose the meaning of data augmentation. In this paper, we propose a limited discriminator GAN (LDGAN) model that explains the results of GAN, which is a model that can not be analyzed externally, such as a black box. The part explained in LDGAN becomes the discriminator model of GAN, and it is possible to check which area of the image is used as the basis for determining fake/real by the discriminator. In the end, a method for limiting the learning of discriminator is proposed based on the described results. Through this, it is possible to avoid the overfitting problem of the discriminator and to generate various images different from the learning image. The LDGAN method allows users to perform meaningful data augmentation with only specific objects except for complex images or backgrounds that require analysis. Compare the LDGAN method with the existing DCGAN and present the extensive simulation results. The extensive simulation result shows that the image generated by the proposed LDGAN including the estimation area is about 10% more.

데이터에 내포되어 있는 주요 정보나 지식을 추출해 내는 기계학습 방법에서 주요 이슈의 하나는 지식 표현방식이다. 여러 가지 구조로 표현될 수 있는 지식을 모델이라고 부른다. 모델에는 그 내부 구조에 따라 트리구조, 네트워크 구조, 리스트 구조, 규칙 등 다양한 구조로 나눈다. 구조의 차이는 단지 표현의 차이뿐만 아니라 그것이 갖는 문제해결 능력에도 차이가 있다. 본 논문에서는 모델을 간략화 시켜 오버피팅 문제를 해결하고 분류 능력을 향상시키는 방법을 제안한다. 모델을 단순화 시키는데 사용되는 파티션 유틸리티 기준함수 제시하고 휴리스틱을 이용하여 균형 잡힌 계층 구조를 생성하는 방법을 제안한다.

One of major issues in machine learning that extracts and acquires knowledge implicit in data is to find an appropriate way of representing it. Knowledge can be represented by a number of structures such as networks, trees, lists, and rules. The differences among these exist not only in their structures but also in effectiveness of the models for their problem solving capability. In this paper, we propose partition utility as a criterion function for clustering that can lead to simplification of the model and thus avoid overfitting problem. In addition, a heuristic is proposed as a way to construct balanced hierarchical models.