<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
    <channel>
        <title>iam_hyun.log</title>
        <link>https://velog.io/</link>
        <description></description>
        <lastBuildDate>Tue, 26 Jul 2022 12:19:10 GMT</lastBuildDate>
        <docs>https://validator.w3.org/feed/docs/rss2.html</docs>
        <generator>https://github.com/jpmonette/feed</generator>
        <copyright>Copyright (C) 2019. iam_hyun.log. All rights reserved.</copyright>
        <atom:link href="https://v2.velog.io/rss/iam_hyun" rel="self" type="application/rss+xml"/>
        <item>
            <title><![CDATA[[1주차] Regression Model]]></title>
            <link>https://velog.io/@iam_hyun/1%EC%A3%BC%EC%B0%A8-Regression-Model</link>
            <guid>https://velog.io/@iam_hyun/1%EC%A3%BC%EC%B0%A8-Regression-Model</guid>
            <pubDate>Tue, 26 Jul 2022 12:19:10 GMT</pubDate>
            <description><![CDATA[<blockquote>
<p>모든 내용은 앤드루 응 교수님의 강의를 듣고 필자가 정리를 위해서 작성하는 것입니다. 오류가 있을 경우 댓글로 알려주세요.</p>
</blockquote>
<h2 id="0-용어-정리">0. 용어 정리</h2>
<p>$Training\ set\ :\ 학습\ 데이터\ 셋$
$x : input\ variable,\ feature$
$y : output\ variable,\ target\ variable$
$m : number\ of\ training\ examples$
$(x, y) : single\ training\ example$
$(x^{(i)}, y^{(i)}) : i^{th} training\ example$
exponent: 거듭제곱</p>
<h2 id="1-linear-regression-model">1. Linear Regression Model</h2>
<p>Regression model은 값을 예측합니다. (ex: 집 크기에 따른 집 가격)
<img src="https://velog.velcdn.com/images/iam_hyun/post/e6a3babe-009d-4e77-a63c-aba5db2ede57/image.png" alt=""></p>
<p>training set을 learning algorithm에 넣고 학습을 하면 &quot;f&quot;라는 model이 나오게 됩니다.
 이러한 f에 x를 넣었을 때의 예측 값을 $\hat{y}$이라고 합니다. <strong>$y$는 training set의 실질적인 값인 반면, $\hat{y}$은 추정 값으로 실제 값과 같을 수도 있고, 아닐 수도 있습니다.</strong></p>
<blockquote>
<p>앞서 말한 모델 f는 다음과 같이 쓸 수 있습니다.</p>
</blockquote>
<p>$f_{w,b}(x) = wx + b$ 이러한 f에 대한 식은 흔히 쓰는 $f(x)$와 같습니다. </p>
]]></description>
        </item>
        <item>
            <title><![CDATA[[1주차] Supervised VS Unsupervised Learning]]></title>
            <link>https://velog.io/@iam_hyun/1%EC%A3%BC%EC%B0%A8-Supervised-VS-Unsupervised-Learning</link>
            <guid>https://velog.io/@iam_hyun/1%EC%A3%BC%EC%B0%A8-Supervised-VS-Unsupervised-Learning</guid>
            <pubDate>Tue, 19 Jul 2022 10:04:56 GMT</pubDate>
            <description><![CDATA[<blockquote>
<p>모든 내용은 앤드루 응 교수님의 강의를 듣고 필자가 정리를 위해서 작성하는 것입니다. 오류가 있을 경우 댓글로 알려주세요.</p>
</blockquote>
<ul>
<li>Supervised Learning</li>
<li>Unsupervised Learning</li>
<li>Recommenders Systems</li>
<li>Reinforcement Learning</li>
</ul>
<p>머신러닝은 크게 두 가지 유형으로 나뉘어집니다. <strong>Supervised Learning</strong> 과 <strong>Unsupervised Learning</strong>으로 나뉘어집니다. </p>
<p>앤드루 응 교수님의 기계학습 특화 과정의 Course 1, 2에서 Supervised Learning과 Unsupervised Learning에 대하여 배우게 되고, Course 3에서 Recommnders Systems와 Reinforcement Learning에 대하여 배우게 됩니다.</p>
<h2 id="1-supervised-learning">1. Supervised Learning</h2>
<p>supervised learning은 input X와 output y과 mapping이 되어있는 학습을 의미합니다.</p>
<table>
<thead>
<tr>
<th>Input(X)</th>
<th>Output(Y)</th>
<th>Application</th>
</tr>
</thead>
<tbody><tr>
<td>email</td>
<td>Spam(0/1)</td>
<td>spam filter</td>
</tr>
<tr>
<td>audio</td>
<td>text transcripts</td>
<td>speech recognition</td>
</tr>
<tr>
<td>English</td>
<td>Spanish</td>
<td>machine translation</td>
</tr>
<tr>
<td>ad, user info</td>
<td>Click(0/1)</td>
<td>online advertising</td>
</tr>
<tr>
<td>image, radar info</td>
<td>position of other cars</td>
<td>self-driving car</td>
</tr>
<tr>
<td>image of phone</td>
<td>defect(0/1)</td>
<td>visual inspection</td>
</tr>
</tbody></table>
<p>위의 표와 같이 input과 output이 mapping된 data가 주어진다는 것이 supervised learning의 특징입니다. 이러한 Supervised learning에는 대표적으로 regression과 classification이 있습니다.</p>
<h3 id="a-regression">A. Regression</h3>
<p>Regression은 가능한 수 많은 output으로 부터 숫자를 예측합니다. 
<img src="https://velog.velcdn.com/images/iam_hyun/post/401b8e5d-f6a8-4697-a18f-2f6271d03ad5/image.png" alt="">
위의 그래프는 집의 면적과 집의 가격을 x모양으로 plot해놓은 그래프입니다. 이 데이터에서는 집의 면적이 input(X), 가격이 output(Y)가 됩니다. 분홍색 곡선 혹은 직선과 같은 선들이 집의 면적에 따른 집의 가격을 예측할 수 있게 해줍니다.</p>
<blockquote>
<p>Supervised Learning part2</p>
</blockquote>
<h3 id="b-classification">B. Classification</h3>
<p>Classification은 Regression과 다르게 가능한 output의 종류가 적을 때 사용합니다. </p>
<p>본 강의에서는 예시로 Breast Cancer Detection을 보여줍니다.
<strong>(Regression에서는 가능한 output이 위의 예시인 집의 가격처럼 여러 숫자가 가능하지만, Classification에서는 아래의 예시와 같이 악성 종양 혹은 양성 종양처럼 output의 종류는 두가지로 그 종류가 적습니다.)</strong>
<img src="https://velog.velcdn.com/images/iam_hyun/post/ae4efdda-0244-41a1-aa72-aa55c3b83a41/image.png" alt="">
이러한 경우 위의 그림 같은 x-y축 그래프보다는, 
<img src="https://velog.velcdn.com/images/iam_hyun/post/a7344072-6ab1-4895-8f37-e8c78121259d/image.png" alt="">
이러한 수직선에 표기하는 것이 더 알아보기 쉽습니다. 이렇게 표기할 경우 여러가지 class 즉 category로 분류된 output을 알 수 있습니다.</p>
<p>이렇듯 Classification 알고리즘은 category를 예측할 수 있게 도와줍니다.</p>
<p><img src="https://velog.velcdn.com/images/iam_hyun/post/bffb6b41-17c9-438a-9262-6d7bcb65a08f/image.png" alt=""></p>
<p>위으 그림 처럼 x-y축에 표시하여 boundary를 지정해주며 category를 구분할 수도 있습니다.</p>
<h3 id="c-recap---regression-vs-classification">C. Recap - Regression VS Classification</h3>
<table>
<thead>
<tr>
<th>Regression</th>
<th>Classification</th>
</tr>
</thead>
<tbody><tr>
<td>값을 예측</td>
<td>category를 예측</td>
</tr>
<tr>
<td>가능한 output이 무한함</td>
<td>가능한 output이 적음</td>
</tr>
</tbody></table>
<h2 id="2-unsupervised-learning">2. Unsupervised Learning</h2>
<p>Unsupervised Learning이란 특정 input에 대하여 정답 output이 주어지는 Supervised Learning과 달리 Unsupervised Learning은 input에 대한 output이 미리 주어지지 않고 인공지능이 input에 대한 상관관계를 알아내는 알고리즘입니다.</p>
<p><img src="https://velog.velcdn.com/images/iam_hyun/post/5f3e1875-b149-42b0-b7b3-2124f1c1f65f/image.png" alt=""></p>
<p>위의 그림과 같이 전체 input에 대하여 clustering을 통해 상관관계를 파악합니다.</p>
<p>Unsupervised Learning의 예로는</p>
<ul>
<li>Clustering</li>
<li>Anomaly detection</li>
<li>Dimensionality reduction
이 있습니다.</li>
</ul>
<h3 id="a-clustering">A. Clustering</h3>
<p>Clustering은 전체 input들 중에서 유사한 data들을 하나의 group으로 지어주는 알고리즘입니다.</p>
<h3 id="b-anomaly-detection">B. Anomaly Detection</h3>
<p>이상 data를 찾는 알고리즘입니다.</p>
<h3 id="c-dimensionality-reduction">C. Dimensionality Reduction</h3>
<p>큰 data를 아주 작게 압축하는 알고리즘입니다.</p>
]]></description>
        </item>
        <item>
            <title><![CDATA[머신러닝 공부를 시작하며...]]></title>
            <link>https://velog.io/@iam_hyun/%EB%A8%B8%EC%8B%A0%EB%9F%AC%EB%8B%9D-%EA%B3%B5%EB%B6%80%EB%A5%BC-%EC%8B%9C%EC%9E%91%ED%95%98%EB%A9%B0</link>
            <guid>https://velog.io/@iam_hyun/%EB%A8%B8%EC%8B%A0%EB%9F%AC%EB%8B%9D-%EA%B3%B5%EB%B6%80%EB%A5%BC-%EC%8B%9C%EC%9E%91%ED%95%98%EB%A9%B0</guid>
            <pubDate>Tue, 19 Jul 2022 09:45:55 GMT</pubDate>
            <description><![CDATA[<p><img src="https://velog.velcdn.com/images/iam_hyun/post/702b91a5-8f96-49c2-84fb-02cfdd56f6f8/image.png" alt=""></p>
<p>필자는 이제 학부 5학기를 수료하였다. 방학 때 뭐를 공부할지 고민하던 중 친한 선배에게 스탠포드 앤드루 응 교수님의 <a href="https://www.coursera.org/specializations/machine-learning-introduction#courses">강의</a>를 추천받았다. </p>
<p>차근 차근 강의를 듣고 각 주차별로 테마 하나(사진 속 &gt;표시) 하나씩 수강하고 글을 쓰고자 한다. 아래 사진은 1주차 테마별 강의이다. <img src="https://velog.velcdn.com/images/iam_hyun/post/5c655817-8f20-47a9-9d77-4fc2a5ff2391/image.png" alt=""> </p>
]]></description>
        </item>
        <item>
            <title><![CDATA[[TEDed] What’s so great about the Great Lakes? - Cheri Dobbs and Jennifer Gabrys]]></title>
            <link>https://velog.io/@iam_hyun/TEDed-Whats-so-great-about-the-Great-Lakes-Cheri-Dobbs-and-Jennifer-Gabrys</link>
            <guid>https://velog.io/@iam_hyun/TEDed-Whats-so-great-about-the-Great-Lakes-Cheri-Dobbs-and-Jennifer-Gabrys</guid>
            <pubDate>Tue, 19 Jul 2022 09:42:17 GMT</pubDate>
            <description><![CDATA[<p>!youtube[gBRcOLcEwF0]</p>
<p>What&#39;s so great about the Great Lakes? They&#39;re known as America&#39;s inland seas. The North American Great Lakes Huron, Ontario,  Michigan, Erie, Superior are so massive that they border eight states and contain 23 quadrillion liters of water. That&#39;s enough to cover the land area of the contiguous United States three meters deep. These vast bodies of water span forest, grassland and wetland habitats, supporting a region that&#39;s home to over 3,500 species. But how did such a vast and unique geological feature come to be? The story begins near the end of the last ice age over 10,000 years ago, a time when the climate was warming and the glaciers that cloaked the Earth&#39;s surface began their slow retreat. Thes immense ice sheets carved out a series of basins. Those basins filled with water as the ice began to melt, creating the world&#39;s largest area of freshwater lakes. Over time, channels developed between these basins, and water began to flow in an ongoing exchange that persists to this day. In fact, today, the interconnected Great Lakes contain almost 20% of the world&#39;s supply of fresh surface water. The water&#39;s journey begins in the far north of Lake Superior, which is the deepest, coldest, and clearest of the lakes, containing half the system&#39;s water. Lake Superior sinks to depths of 406 meters, creating a unique and diverse ecosystem that includes more than 80 fish species. A given drop of water spends on average 200 years in this lake before flowing into Lake Michigan of Lake Huron. Linked by the Straits of Mackinac, these two lakes are technically one. To the west lies Lake Michigan, the third largest of the lakes by surface area. Water slowly moves through its cul-de-sac shape and encounters the world&#39;s largest freshwater dunes, many wildlife species, and unique fossilized coral. To the east is Lake Huron, which has the longest shoreline. It&#39;s sparsely populated, but heavily forested, including 7,000-year-old petrified trees. Below them, water continues to flow southeastwards from Lake Huron into Lake Erie. This lake&#39;s status as the warmest and shallowest of the five has ensured an abundance of animal life, including millions of migrating birds. Finally, the water reaches its last stop by dramatically plunging more than 50 meters down the thundering Niagara Falls into Lake Ontario, the smallest lake by surface area. From there, some of this well-traveled water enters the St. Lawrence River, eventually reaching the Atlantic Ocean. In addition to being a natural wonder, the perpetually flowing Great Lakes brings us multiple benefits. They provide natural water filtration, flood control, and nutrient cycling. By moving water across more than 3,200 kilometers, the Great Lakes also provide drinking water for upward of 40 million people and 212 billion liters a day for the industries and farms that line their banks. But our dependence on the system is having a range of negative impacts, too. The Great Lakes coastal habitats are being degraded and increasingly populated, exposing the once pristine waters to industrial, urban, and agricultural pollutants. Because less than 1% of the water leaves the Lake&#39;s system annually, decades-old pollutants still lurk in its waters. Humans have also inadvertently introduced more than 100 non-native and invasive species into the lakes, such as zebra and quagga mussels, and sea lampreys that have decimated some indigenous fish populations. On a larger scale, climate change is causing the waters to warm, thus reducing water levels and changing the distribution of aquatic life. Luckily, in recent years, governments have started to recognize the immense value of this natural resource. Partnerships between the United States and Canada are underway to reduce pollution, protect coastal habitats, and halt the spread of invasive species. Protecting something as massive as the Great Lakes system will require the collaboration of many organizations, but the effort is critical if we can preserve the wonder of this flowing inland sea.</p>
]]></description>
        </item>
        <item>
            <title><![CDATA[다시 시작.]]></title>
            <link>https://velog.io/@iam_hyun/%EB%8B%A4%EC%8B%9C-%EC%8B%9C%EC%9E%91</link>
            <guid>https://velog.io/@iam_hyun/%EB%8B%A4%EC%8B%9C-%EC%8B%9C%EC%9E%91</guid>
            <pubDate>Tue, 19 Jul 2022 09:39:38 GMT</pubDate>
            <description><![CDATA[<p>티스토리, Jekyll, Hexo 여러 블로그를 사용해보다 이제 마음 다 잡고 새롭게 블로그를 시작해보려고 한다.</p>
<p>07.19.22. 동방에서...</p>
]]></description>
        </item>
    </channel>
</rss>