Previous post is about the supervised learning trained from labeled data. But many cases, we afford to have labeled data set to train.

The other way to make your model for your data is Unsupervised learning trained from unlabeled data.

If your learning algorithm have some problems, How can we detect and solve it?

Here is the solution. The one of the most important parts to train and build machine learning is debugging your algorithm. Debugging can give you the insight for your next step. Without debugging, you don’t sure your algorithm is trained well without overfitting or underfitting.

This post is related to the previous post, SVM. This post is about the advanced way to solve non-linear problem and choose proper hyper paramters for SVM.

If your data set is linear , it is enough to use linear classifcation like (SVM, Logistic regression)

Unless, you can’t get the fine model for your dataset.

Here is a solution, the kernel. In this post, i will describe how kernel can solve non-linear problems.

Virtual Container로 유명한 도커의 특징과 실제로 환경 구축한 내용을 정리해보았다.

• Ubuntu 18.04 LTS 64bits

서버 업그레이드 후, 개발 환경 설정을 위하여 텐서플로우를 설치하려고 봤더니, 이전과는 버젼이 많이 달라져서 새로 개발 환경 설정 방법에 대해 정리해보고자 한다.

In this post, I will introduce about the SVM(Support vector machine)

SVM is one of the powerful algorithms for classification.

I will compare with the logistic regression the most common algorithm for classification to make you know what the differences are.

The origins of Neural network is a try to mimic the human’s brain.

In this post, i will introduce how Neural network works and how to implement.

What is over fitting and under fitting? How can we avoid this problem?