<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:media="http://search.yahoo.com/mrss/" xmlns:ynews="http://news.yahoo.com/rss/">
    <channel>
        <title>Nova Reader - Subject</title>
        <link>https://www.novareader.co</link>
        <description>Default RSS Feed</description>
        <language>en-us</language>
        <copyright>Newgen KnowledgeWorks</copyright>
        <item>
            <title><![CDATA[A message-passing multi-task architecture for the implicit event and polarity detection]]></title>
            <media:thumbnail url="https://storage.googleapis.com/nova-demo-unsecured-files/unsecured/content-1765977616159-7976fa35-8af8-4d8c-a3b8-5a5d63782bed/cover.png"></media:thumbnail>
            <link>https://www.novareader.co/book/isbn/10.1371/journal.pone.0247704</link>
            <description><![CDATA[<p class="para" id="N65539">Implicit sentiment analysis is a challenging task because the sentiment of a text is expressed in a connotative manner. To tackle this problem, we propose to use textual events as a knowledge source to enrich network representations. To consider task interactions, we present a novel lightweight joint learning paradigm that can pass task-related messages between tasks during training iterations. This is distinct from previous methods that involve multi-task learning by simple parameter sharing. Besides, a human-annotated corpus with implicit sentiment labels and event labels is scarce, which hinders practical applications of deep neural models. Therefore, we further investigate a back-translation approach to expand training instances. Experiment results on a public benchmark demonstrate the effectiveness of both the proposed multi-task architecture and data augmentation strategy.</p>]]></description>
            <pubDate><![CDATA[2021-03-01T00:00]]></pubDate>
        </item><item>
            <title><![CDATA[The Huawei and Snowden Questions]]></title>
            <media:thumbnail url="https://storage.googleapis.com/nova-demo-unsecured-files/unsecured/content-1764505573692-f76c40a5-84b8-4351-992f-7665efa4d91a/9783319749501.png"></media:thumbnail>
            <link>https://www.novareader.co/book/isbn/9783319749501</link>
            <description><![CDATA[This open access book answers two central questions: firstly, is it at all possible to verify electronic equipment procured from untrusted vendors? Secondly, can I build trust into my products in such a way that I support verification by untrusting customers?  In separate chapters the book takes readers through the state of the art in fields of computer science that can shed light on these questions. In a concluding chapter it discusses realistic ways forward.  

In discussions on cyber security, there is a tacit assumption that the manufacturer of equipment will collaborate with the user of the equipment to stop third-party wrongdoers. The Snowden files and recent deliberations on the use of Chinese equipment in the critical infrastructures of western countries have changed this. The discourse in both cases revolves around what malevolent manufacturers can do to harm their own customers, and the importance of the matter is on par with questions of national security.

This book is of great interest to ICT and security professionals who need a clear understanding of the two questions posed in the subtitle, and to decision-makers in industry, national bodies and nation states.]]></description>
            <pubDate><![CDATA[2019-02-18T18:30]]></pubDate>
        </item>
    </channel>
</rss>