web index edit

web/index.html

@@ -151,8 +151,8 @@ <h2>Image Generation</h2>
                 <p>1. 아래에 보이는 100개의 픽셀을 <b>z</b>의 각 숫자를 나타냅니다.<br/>2. 만들어진 사진 위에 마우스를 올리면 사진에 사용된 <b>z</b>가 보입니다.<br/>3. 만들어진 이미지를 누르시면 그 이미지의 <b>z</b>가 복사됩니다.</p>
             </div>
             <div lang="en" class="col-md-8 col-md-offset-2 col-xs-12 text-left">
-                <br/><p><b>Neural Face</b> uses a vector <b>z</b> that consists of 100 real numbers range from 0 to 1.</p>
-                <p>1. Each pixel in bellow pallete represents each values in <b>z</b>.<br/>2. If you hover your mouse over an image, <b>z</b> for that image will be displayed.<br/>3. If you click an image, <b>z</b> will be copyed to the palette.</p>
+                <br/><p><b>Neural Face</b> uses a vector <b>z</b> that consists of 100 real numbers ranging from 0 to 1.</p>
+                <p>1. Each pixel in the below pallete represents a value in <b>z</b>.<br/>2. If you hover your mouse over an image, <b>z</b> for that image will be displayed.<br/>3. If you click an image, <b>z</b> will be copied to the palette.</p>
             </div>
             <div class="col-md-8 col-md-offset-2 col-xs-12">
                 <p lang="ko"><small>(브라우저 성능에 따라 1~10초가 걸립니다)</small></p>
@@ -236,7 +236,7 @@ <h2 lang="en">Algorithm</h2>
             <div lang="en" class="col-md-8 col-md-offset-2 col-xs-12 text-left">
                 <p><a href="http://arxiv.org/pdf/1511.06434v2.pdf" target="_blank">DCGAN</a>, which is the core of <b>Neural Face</b>, consists of two different neural networks which are:</p>
                 <p>1. <b>Generator (G)</b> that generates an image<br/>2. <b>Discriminator (D)</b> that discriminate real images from generated images</p>
-                <p>Two neural networks read numerous of images to deceive each other and this kind of learning is called <a href="https://en.wikipedia.org/wiki/Adversarial_machine_learning">Adversarial Learning</a>. Because of this, <b>Generator</b> and <b>Discriminator</b> are described as a <b>theif</b> and a <b>police</b>.</p>
+                <p>Two neural networks compete as one tries to deceive the other. This kind of learning is called <a href="https://en.wikipedia.org/wiki/Adversarial_machine_learning">Adversarial Learning</a>. Because of this, <b>Generator</b> and <b>Discriminator</b> are described as a <b>thief</b> and <b>police</b>, respectively.</p>
             </div>
             <br/>
             <img class="col-md-6 col-md-offset-3 col-xs-10 col-xs-offset-1" src="img/model.png">
@@ -246,8 +246,8 @@ <h2 lang="en">Algorithm</h2>
                 <p>구분자는 인공 신경망에 실제 이미지를 넣은 결과를 <b>1</b>로, 만들어진 이미지의 결과는 <b>0</b>으로 구분하도록 학습합니다. 반대로 생성자는 Gaussian Distribution을 따르는 <b>z</b>라는 확률 변수를 두고, 사람의 이미지의 확률 분포를 <b>z</b>를 사용해 계산합니다. 이렇게 만들어진 이미지를 구분자가 실제 이미지라고 잘못 판단하도록 계속 학습합니다.</p>
             </div>
             <div lang="en" class="col-md-8 col-md-offset-2 col-xs-12 text-left">
-                <p><br/>Generator and Discriminator consists of <a href="https://www.quora.com/How-does-a-deconvolutional-neural-network-work" target="_blank">Deconvolutional Network (DNN)</a> and <a href="http://cs231n.github.io/convolutional-networks/" target="_blank">Convolutional Neural Network (CNN)</a>. <b>CNN</b> is a neural network which encode the hundreds of pixels of an image into a vector of small dimensions (<b>z</b>) which is a summary of the image and <b>DNN</b> is a network that learns filters to recover the original image from <b>z</b>.</p>
-                <p>When a real image is given, Discirminator should tell <b>1</b> and <b>0</b> for fake generated image from Generator. In the contrast, Generator generate an image from <b>z</b> which follows Gaussian Distribution and try to figure out the distribution of human image from <b>z</b>. In this way, a Generator tries to cheat Discriminator into making a wrong decision.</p>
+                <p><br/>Generator and Discriminator consist of <a href="https://www.quora.com/How-does-a-deconvolutional-neural-network-work" target="_blank">Deconvolutional Network (DNN)</a> and <a href="http://cs231n.github.io/convolutional-networks/" target="_blank">Convolutional Neural Network (CNN)</a>. <b>CNN</b> is a neural network which encodes the hundreds of pixels of an image into a vector of small dimensions (<b>z</b>) which is a summary of the image. <b>DNN</b> is a network that learns filters to recover the original image from <b>z</b>.</p>
+                <p>When a real image is given, Discriminator should output <b>1</b> or <b>0</b> for whether the image was generated from Generator. In the contrast, Generator generates an image from <b>z</b>, which follows a Gaussian Distribution, and tries to figure out the distribution of human images from <b>z</b>. In this way, a Generator tries to cheat Discriminator into making a wrong decision.</p>
             </div>
         </div>
     </section>
@@ -258,7 +258,7 @@ <h2>Results</h2>
             <div class="col-md-8 col-md-offset-2 col-xs-12 text-left">
                 <p lang="ko"><b>프사 뉴럴</b>를 학습시키기 위해 인터넷에 10만 개 이상의 사진들을 모았고 이 사진들에서 얼굴 사진만 잘라서 얼굴 데이터 셋을 만들었습니다. 코드는 최근에 구글에서 공개한 <a href="https://www.tensorflow.org/" target="_blank">TensorFlow</a>로 구현했으며 GTX 980 Ti를 사용하여 이틀간 학습시켰습니다.</p>
                 <p lang="ko">아래는 초기 학습 단계에서 프사 뉴럴이 정해진 <b>z</b>로 얼굴 사진을 만들어 가는 과정을 보여줍니다.</p>
-                <p lang="en">More than 100K images are crawled from online communities and those images are cropped by using <a href="https://github.com/cmusatyalab/openface">openface</a> which is a face recognition framework. <b>Neural Face</b> is implemented with <a href="https://www.tensorflow.org/" target="_blank">TensorFlow</a> and GTX 980 Ti is used to train for two days.</p>
+                <p lang="en">More than 100K images are crawled from online communities and those images are cropped by using <a href="https://github.com/cmusatyalab/openface">openface</a> which is a face recognition framework. <b>Neural Face</b> is implemented with <a href="https://www.tensorflow.org/" target="_blank">TensorFlow</a> and a GTX 980 Ti is used to train for two days.</p>
                 <p lang="en">Below is a series of images generated by <b>Generator</b> with a fixed <b>z</b> between the first and the fith epoch of training.</p>
                 <video autoplay loop muted class="col-md-6 col-md-offset-3 col-xs-10 col-xs-offset-1">
                     <source src="videos/training.mp4" type="video/mp4">
@@ -275,7 +275,7 @@ <h2>Results</h2>
 
             <div class="col-md-8 col-md-offset-2 col-xs-12 text-left">
                 <p lang="ko"><br/><br/>평균값 <b>0</b>에서 랜덤한 차원의 값을 조금씩 바꾸면 아래와 같은 변화를 볼 수 있습니다.</p>
-                <p lang="en"><br/><br/>Below images are genearted by changing the values of <b>z</b> continuously, starting from the average value (<b>0</b>) to -1 or 1.</p>
+                <p lang="en"><br/><br/>The below images are generated by changing the values of <b>z</b> continuously, starting from the average value (<b>0</b>) to -1 or 1.</p>
             </div>
             <div class="col-md-8 col-md-offset-2 col-xs-12">
                 <video autoplay loop muted class="col-md-6 col-md-offset-3 col-xs-10 col-xs-offset-1">
@@ -285,7 +285,7 @@ <h2>Results</h2>
 
             <div class="col-md-8 col-md-offset-2 col-xs-12 text-left">
                 <p lang="ko"><br/><br/>아래의 사진들은 100차원의 <b>z</b> 값 중에서 임의의 차원들을 -1부터 1까지 바꾸면서 <b>생성자</b> 신경망에 넣은 결과이며, 점점 미소를 짓거나, 안경이 생기거나, 흑백 사진이 되거나, 성별이 바뀌는 등의 결과를 확인하실 수 있습니다.</p>
-                <p lang="en"><br/><br/>Below images are generated by changing ten different values of <b>z</b> from -1 to 1. People in image become smiling, wearing glasses, turing into black and white images, and chaning in to different sex.</p>
+                <p lang="en"><br/><br/>The below images are generated by changing ten different values of <b>z</b> from -1 to 1. People in the images vary in characteristics such as smiling, wearing glasses, turning into black and white images, and changing into different sex.</p>
             </div>
         </div>
         <div class="row">