Terryhung

\section{Experiment}
Our fine-tuning takes an already learned model: BVLC CaffeNet Model. CaffeNet is modified by AlexNet. This model is the result of Caffnet training on ImageNet. We set the result of fine-tuning as our baseline.

We use the dataset from Microsoft: Clickture-FilteredDog. This is a subset of the Clickture-Full dataset which only contains the dog breed related items. We pick out 107 class of this subset which contains more than 100 images total 89,910 images. We use 5-fold to split this dataset: 7,1932 images for training and 17,978 for testing.

Our result on Clickture-FilteredDog in Table 1 and Table 2. Our network achieves accuracy of \textbf{50.5\%}. The best performance with fine-tuning is 46.2\%.

In Table 1, the result of our first approach, average vector, does not exceed baseline(fine-tuning). Our MMD loss does not fall down after 7000? iterations, so we consider that average reduces some information in text such that the performance do not better than fine-tuning. Form t-SNE algorith

Terryhung

\begin{table}[t!]
	\centering
	\caption{Example of correct and fail cases.}
	\label{table:dog_compare}
	\begin{adjustbox}{max width=90mm}
	\begin{tabular}{|l|>{\centering\arraybackslash}m{2in}|}
		\hline
		 &  \quad \quad Border Collie \newline  \raisebox{-\totalheight}{\includegraphics[scale=0.2]{correction_border_collie}} \vfill \\ \hline

		baseline & \quad \quad 	Australian Shepherd \newline \raisebox{-\totalheight}{\includegraphics[scale=0.2]{correction_australian_shepherd_1}} \vfill \\ \hline

Terryhung

//Java
public static String ParsingHost() throws Exception {
    String url = "http://www.cna.com.tw/news/aloc/201710030325-1.aspx";
    URL aURL = new URL(url);
    String host = aURL.getHost();
    return host;
}        

Terryhung

// JAVA

    public static void parsing() throws exception {
        string url = "http://www.cna.com.tw/news/aloc/201710030325-1.aspx";
        document doc = jsoup.connect(url)
            .header("user-agent", "mozilla/5.0 (linux; android 7.0; samsung sm-g950u build/nrd90m) applewebkit/537.36 (khtml, like gecko) samsungbrowser/5.2 chrome/51.0.2704.106 mobile safari/537.36")
            .get();

        string target = "div.news_article";
        string remove = "script, button";

Terryhung

//JAVA
        public static void AddDocHead(Elements doc) throws Exception {
        Element head = doc.first();
        head.prepend("<style>.article-header{content: '';background: linear-gradient(to top, #000 1rem,#fff 40rem,#fff 100%);height: 35rem;display: block;}.main-image{width: 100%;max-height: 700px;opacty: 0.9}.article-content{background-color: #fff;position: absolute;width: 90%;z-index: 1;margin-left: 4%; margin-right: 5%;margin-top: -2rem;font-size: 25px; padding-left: 5px; padding-right: 5px}.article-header:after {content: '';background: #000;background: -moz-linear-gradient(top, #000 28%, #fff 58%, #fff 100%);background: -webkit-linear-gradient(top, #000 22rem,#fff 40rem,#fff 100%);background: linear-gradient(to bottom, #000 22rem,#fff 40rem,#fff 100%);height: 40rem;display: block;} .article-content img{width: 100%}</style>");
    }

    public static void Parsing() throws Exception {
        String url = "http://www.cna.com.tw/news/aloc/201710030325-1.aspx";
        Document doc = Jsoup.connec

Terryhung

// Java
    //without image url
    public static void AddDocHead(Elements doc) throws Exception {
        Element head = doc.first();
        head.prepend("<style>.article-header{content: '';background: linear-gradient(to top, #000 1rem,#fff 40rem,#fff 100%);height: 5rem;display: block;}.main-image{width: 100%;max-height: 700px;opacty: 0.9}.article-content{background-color: #fff;position: absolute;width: 90%;z-index: 1;margin-left: 4%; margin-right: 5%;margin-top: -2rem;font-size: 25px; padding-left: 5px; padding-right: 5px}.article-header:after {content: '';background: #000;background: -moz-linear-gradient(top, #000 28%, #fff 58%, #fff 100%);background: -webkit-linear-gradient(top, #000 22rem,#fff 40rem,#fff 100%);background: linear-gradient(to bottom, #000 22rem,#fff 40rem,#fff 100%);height: 40rem;display: block;} .article-content img{width: 100%} div{white-space: pre-wrap; white-space: -moz-pre-wrap; word-wrap: break-word;}</style>");
    }

    // with image url
    public static void AddDocHead(Elements

Terryhung

    public static String AddDocHead(String s) throws Exception {

        String _s = "<style> .article{padding-left: 18px;padding-right: 18px;padding-top: 15px;word-break: break-all}.article-header{font-size: 27px;color: #1a1a1a}.article-info{font-size: 16px;color: #c2c2c2;margin-top: 18px;margin-bottom: 18px}.article-content{margin-top: 18px;font-size: 22px;color: #818181}.main-image{width: 100%;max-height: 500px;margin-top: 20px;margin-bottom: 20px}hr{margin-top: 20px;color: #f5f5f5} img{width: 100%}</style>".concat(s);

        return _s;
    }

    public static void Parsing() throws Exception {
        String url = "http://www.cna.com.tw/news/aloc/201710030325-1.aspx";
        Document doc = Jsoup.connect(url)

Terryhung

def lower_df(df):
    pattern = re.compile('\W+')
    for field in ['name', 'category_name', 'brand_name', 'item_description']:
        _str = list(map(
            lambda x: re.sub(pattern, ' ', x),
            df[field].str.lower()
        ))
        df[field] = _str

Terryhung

    def forward(self, feature):
        for f in feature:
            f = torch.from_numpy(f)
            f = f.float()
            lstm_out, self.hidden = self.lstm(
                f.view(1, 1, len(f)), self.hidden)

        tag_space = self.hidden2tag(lstm_out.view(1, self.hidden_dim))
        tag_scores = F.log_softmax(tag_space, dim=1)

Terryhung

class Solution:
    # 0<= value <= 100  
    base = 101
    # leetcode popular prime number
    p = 10 ** 9 + 7
    
    def findLength(self, nums1: List[int], nums2: List[int]) -> int:
        l = 0
        r = min(len(nums1), len(nums2))