<div class="contents_title"> <div style="float:left">Contents</div> <div class="contentspagination"> <div class="pagination PagedList-pager"><ul><li class="previous PagedList-skipToFirst"><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=1" title="Show first page">&laquo;</a></li><li class="PagedList-skipToPrevious"><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=4" title="Show previous page">&lsaquo;</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=1" title="Show page 1">1</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=2" title="Show page 2">2</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=3" title="Show page 3">3</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=4" title="Show page 4">4</a></li><li class="active"><a>5</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=6" title="Show page 6">6</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=7" title="Show page 7">7</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=8" title="Show page 8">8</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=9" title="Show page 9">9</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=10" title="Show page 10">10</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=11" title="Show page 11">11</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=12" title="Show page 12">12</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=13" title="Show page 13">13</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=14" title="Show page 14">14</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=15" title="Show page 15">15</a></li><li class="disabled PagedList-ellipses"><a>&#8230;</a></li><li class="PagedList-skipToNext"><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=6" title="Show next page">&rsaquo;</a></li><li class="next PagedList-skipToLast"><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=17" title="Show last page">&raquo;</a></li></ul></div> </div> </div> <div class="contents" id="volumearticles" start="101"> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54962">Dividing the Ontology Alignment Task with Semantic Embeddings and Logic-Based Modules</a></div> <label>Authors</label> <div class="value authors">Ernesto Jim&#233;nez-Ruiz, Asan Agibetov, Jiaoyan Chen, Matthias Samwald, Valerie Cross</div> <label>Pages</label> <div class="value pages">784 - 791</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200167</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>Large ontologies still pose serious challenges to state-of-the-art ontology alignment systems. In this paper we present an approach that combines a neural embedding model and logic-based modules to accurately divide an input ontology matching task into smaller and more tractable matching (sub)tasks. We have conducted a comprehensive evaluation using the datasets of the Ontology Alignment Evaluation Initiative. The results are encouraging and suggest that the proposed method is adequate in practice and can be integrated within the workflow of systems unable to cope with very large ontologies.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54962" method="post"> <input type="hidden" name="id" value="54962" /> <div id='downloadlink54962' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54962').click(function () { $('form#downloadform54962').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54963">Rewrite or Not Rewrite? ML-Based Algorithm Selection for Datalog Query Answering on Knowledge Graphs</a></div> <label>Authors</label> <div class="value authors">Unmesh Joshi, Ceriel Jacobs, Jacopo Urbani</div> <label>Pages</label> <div class="value pages">792 - 799</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200168</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>Query-driven reasoning techniques with Datalog rules, like Magic Sets (MS), are ideal for implementing query answering on Knowledge Graphs (KGs). For some queries, executing a rewriting procedure like MS is the best choice, but for others a non-rewriting procedure like Query-subquery (QSQ) can be faster. Choosing beforehand which procedure should be used is not trivial and mistakes can be costly. To address this problem, we describe a first-of-its-kind method that builds a Machine Learning (ML) model to predict whether a query should be answered with MS or with QSQ. Experiments on several well-known KGs show that our method can return accurate predictions, and this leads to a significant reduction of the response time of query answering.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54963" method="post"> <input type="hidden" name="id" value="54963" /> <div id='downloadlink54963' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54963').click(function () { $('form#downloadform54963').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54964">Necessary and Sufficient Conditions for Actual Root Causes</a></div> <label>Authors</label> <div class="value authors">Shakil M. Khan, Mikhail Soutchanski</div> <label>Pages</label> <div class="value pages">800 - 808</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200169</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>Reasoning about actual causes of an observed effect is fundamental to many applications. Batusov and Soutchanski (2018) recently presented a first-order logic approach to compute actual causes. Built on a formal theory of action and change, namely the situation calculus, their approach is quite expressive, as it can be used to determine the causes of quantified effects. However, their approach does not find causes from a counterfactual perspective, nor does it link with the regularity approach to causation. This paper proposes a new analysis of actual achievement causes in the situation calculus. We study the natural properties that are necessary for actual causes and conditions that are sufficient for the achievement of an observed (possibly quantified) effect. We identify a property that is both necessary and sufficient for actual achievement causes. This is one of our main contributions. Our discussion leads to a new definition of actual achievement causes that includes the root cause together with a chain of relevant events. We show when our definition is closely related to the recent one proposed by Batusov and Soutchanski (2018).</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54964" method="post"> <input type="hidden" name="id" value="54964" /> <div id='downloadlink54964' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54964').click(function () { $('form#downloadform54964').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54965">Agent Abstraction via Forgetting in the Situation Calculus</a></div> <label>Authors</label> <div class="value authors">Kailun Luo, Yongmei Liu, Yves Lesp&#233;rance, Ziliang Lin</div> <label>Pages</label> <div class="value pages">809 - 816</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200170</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>In an earlier paper, Banihashemi et al. proposed a general framework for agent abstraction based on the situation calculus. They used basic action theories (BATs) to represent agents’ behavior, and mappings to specify how high-level BATs relate to low-level ones. They then defined the concepts of sound/complete abstractions of BATs based on the notion of bisimulation between high-level and low-level models. However, they didn’t address the issue of the construction of an abstraction from a low-level action theory when given a mapping. It turns out that their concept of abstraction is closely related to the well-explored notion of forgetting. In this paper, we explore agent abstraction via forgetting. Firstly, we show that a correct (i.e., sound and complete) abstraction can be obtained via forgetting low-level symbols from the low-level action theory together with axioms for bisimulation. Secondly, we show how to compute via forgetting a correct abstraction in the form of a generalized BAT (i.e., where the initial database, action preconditions and successor state descriptions can be second-order formulas) under a suitable Markovian restriction. Finally, we show that in the propositional case, under the suitable Markovian restriction, correct abstractions are always computable.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54965" method="post"> <input type="hidden" name="id" value="54965" /> <div id='downloadlink54965' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54965').click(function () { $('form#downloadform54965').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54966">BTDE: Block Term Decomposition Embedding for Link Prediction in Knowledge Graph</a></div> <label>Authors</label> <div class="value authors">Tao Luo, Yifan Wei, Mei Yu, Xuewei Li, Mankun Zhao, Tianyi Xu, Jian Yu, Jie Gao, Ruiguo Yu</div> <label>Pages</label> <div class="value pages">817 - 824</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200171</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>Link prediction is the main task of knowledge graph completion, predicting missing relations between entities based the existing links among the entities. The problem of knowledge graph completion can be framed as a third-order binary tensor completion problem. In this case, tensor decomposition seems like a natural solution. And many previous studies have shown that tensor decomposition methods are superior to Trans-based methods in link prediction experiments. Typical tensor decomposition methods are Canonical Polyadic (CP) decomposition and Tucker decomposition. In this paper, we propose Block term decomposition Embedding model (BTDE) for link prediction based on Block term decomposition (which can be seen as a combination of CP decomposition and Tucker decomposition) of the binary tensor representation of knowledge graph triples. The embeddings learned through BTDE is interpretable. In addition, we prove BTDE is fully expressive and derive the bound on its entity and relation embedding dimensionality for full expressivity which is the same as TuckER and smaller than the bound of previous start-of-the-art models ComplEx and SimplE. We show empirically that BTDE outperforms most previous state-of-the-art models across five standard link prediction datasets.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54966" method="post"> <input type="hidden" name="id" value="54966" /> <div id='downloadlink54966' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54966').click(function () { $('form#downloadform54966').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54967">Auction Description Language (ADL): General Framework for Representing Auction-Based Markets</a></div> <label>Authors</label> <div class="value authors">Munyque Mittelmann, Laurent Perrussel</div> <label>Pages</label> <div class="value pages">825 - 832</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200172</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>The goal of this paper is to propose a language for representing and reasoning about the rules governing an auction-based market. Such language is at first interest as long as we want to build up digital market places based on auction, a widely used framework for automated transactions. Auctions may differ in several aspects: single or double-side, ascending or descending, single or multi-unit, open cry or sealed-bid, and so on. This variety prevents an agent to easily switch between different (auction-based) markets. The first requirement for building such agents is to have a general language for describing auction-based markets. Second, this language should also allow the reasoning about the key issues of a specific market, namely the allocation and payment rules. To do so, we define a language in the spirit of the Game Description Language (GDL): the Auction Description Language (ADL) is the first language for describing auctions in a logical framework. In this paper, we illustrate this general dimension by representing two different types of well-known auctions: an English Auction and a Multi-Unit Vickrey Auction. We show the benefit of ADL by deriving properties about these two auction protocols. It also enables us to show in an explicit way what should be assumed about the behavior of a rational bidder.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54967" method="post"> <input type="hidden" name="id" value="54967" /> <div id='downloadlink54967' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54967').click(function () { $('form#downloadform54967').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54968">Multi-Partition Embedding Interaction with Block Term Format for Knowledge Graph Completion</a></div> <label>Authors</label> <div class="value authors">Hung Nghiep Tran, Atsuhiro Takasu</div> <label>Pages</label> <div class="value pages">833 - 840</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200173</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>Knowledge graph completion is an important task that aims to predict the missing relational link between entities. Knowledge graph embedding methods perform this task by representing entities and relations as embedding vectors and modeling their interactions to compute the matching score of each triple. Previous work has usually treated each embedding as a whole and has modeled the interactions between these whole embeddings, potentially making the model excessively expensive or requiring specially designed interaction mechanisms. In this work, we propose the multi-partition embedding interaction (MEI) model with block term format to systematically address this problem. MEI divides each embedding into a multi-partition vector to efficiently restrict the interactions. Each local interaction is modeled with the Tucker tensor format and the full interaction is modeled with the block term tensor format, enabling MEI to control the trade-off between expressiveness and computational cost, learn the interaction mechanisms from data automatically, and achieve state-of-the-art performance on the link prediction task. In addition, we theoretically study the parameter efficiency problem and derive a simple empirically verified criterion for optimal parameter trade-off. We also apply the framework of MEI to provide a new generalized explanation for several specially designed interaction mechanisms in previous models.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54968" method="post"> <input type="hidden" name="id" value="54968" /> <div id='downloadlink54968' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54968').click(function () { $('form#downloadform54968').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54969">Strong Refinements for Hard Problems in Argumentation Dynamics</a></div> <label>Authors</label> <div class="value authors">Andreas Niskanen, Matti J&#228;rvisalo</div> <label>Pages</label> <div class="value pages">841 - 848</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200174</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>Going beyond the more classically studied reasoning problems over argumentation frameworks (AFs), the study of dynamics in argumentation gives rise to new types of computational challenges. This work studies ways of extending the scalability of computational approaches to reasoning about dynamics of abstract argumentation frameworks. In particular, we focus on three recently proposed optimization problems underlying AF dynamics—two variants of enforcement in abstract argumentation and the synthesis of argumentation frameworks from examples—for semantics under which the problems are (presumably) complete for the second level of the polynomial hierarchy. As the main contributions, we show that by bridging recent theoretical results on the persistence of extensions under changes to the structure of AFs with Boolean satisfiability (SAT) counterexample-guided abstraction refinement algorithms for the considered problems, the scalability of state-of-the-art practical algorithms for each of the three problems can be significantly improved.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54969" method="post"> <input type="hidden" name="id" value="54969" /> <div id='downloadlink54969' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54969').click(function () { $('form#downloadform54969').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54970">Algorithms for Dynamic Argumentation Frameworks: An Incremental SAT-Based Approach</a></div> <label>Authors</label> <div class="value authors">Andreas Niskanen, Matti J&#228;rvisalo</div> <label>Pages</label> <div class="value pages">849 - 856</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200175</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>Motivated by the fact that argumentation is intrinsically a dynamic process, the study of representational and computational aspects of dynamics in argumentation is starting to gain more traction. This is also witnessed by the most recent 2019 edition of the International Competition on Computational Models of Argumentation (ICCMA 2019), which introduced a new track focusing on dynamic argumentation frameworks. In this paper, we present an efficient Boolean satisfiability (SAT) based approach to reasoning over dynamic argumentation frameworks. In particular, based on employing incremental SAT solving, we detail algorithms covering all of the reasoning tasks—credulous and skeptical acceptance, as well as the computation of a single and all extensions—and semantics—complete, preferred, stable, and grounded—constituting the ICCMA 2019 dynamic track. Furthermore, we demonstrate empirically that an implementation of the approach is highly competitive.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54970" method="post"> <input type="hidden" name="id" value="54970" /> <div id='downloadlink54970' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54970').click(function () { $('form#downloadform54970').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54971">On Measuring Inconsistency in Relational Databases with Denial Constraints</a></div> <label>Authors</label> <div class="value authors">Francesco Parisi, John Grant</div> <label>Pages</label> <div class="value pages">857 - 864</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200176</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>Real-world databases are often inconsistent. Although there has been an extensive body of work on handling inconsistency, little work has been done on measuring inconsistency in databases. In this paper, building on work done on measuring inconsistency in propositional knowledge bases, we explore inconsistency measures (IMs) for databases with denial constraints. We first introduce new database IMs that are inspired by well-established methods to quantify inconsistency in propositional knowledge bases, but are tailored to the relational database context where data are generally the reason for inconsistency, not the integrity constraints. Then, we analyze the compliance of the database IMs with rationality postulates, and investigate the complexity of the inconsistency measurement problem as well as of the problems of deciding whether the inconsistency is lower than, greater than, or equal to a given threshold.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54971" method="post"> <input type="hidden" name="id" value="54971" /> <div id='downloadlink54971' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54971').click(function () { $('form#downloadform54971').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54972">Abstract Argumentation with Markov Networks</a></div> <label>Authors</label> <div class="value authors">Nico Potyka</div> <label>Pages</label> <div class="value pages">865 - 872</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200177</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>We explain how abstract argumentation problems can be encoded as Markov networks. From a computational perspective, this allows reducing argumentation tasks like finding labellings or deciding credulous and sceptical acceptance to probabilistic inference tasks in Markov networks. From a semantical perspective, the resulting probabilistic argumentation models are interesting in their own right. In particular, they satisfy several of the properties proposed for epistemic probabilistic argumentation by Hunter and Thimm. We also consider an extension to frameworks with deductive support and show that it maintains many of the interesting guarantees of both approaches.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54972" method="post"> <input type="hidden" name="id" value="54972" /> <div id='downloadlink54972' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54972').click(function () { $('form#downloadform54972').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54973">Reuse, Reduce and Recycle: Optimizing Reiter’s HS-Tree for Sequential Diagnosis</a></div> <label>Authors</label> <div class="value authors">Patrick Rodler</div> <label>Pages</label> <div class="value pages">873 - 880</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200178</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>Reiter’s HS-Tree is one of the most popular diagnostic search algorithms due to its desirable properties and general applicability. In sequential diagnosis, where the addressed diagnosis problem is subject to successive change through the acquisition of additional knowledge about the diagnosed system, HS-Tree is used in a stateless fashion. That is, the existing search tree is discarded when new knowledge is obtained, albeit often large parts of the tree are still relevant and have to be rebuilt in the next iteration, involving redundant operations and costly reasoner calls. As a remedy, we propose DynamicHS, a variant of HS-Tree that avoids these redundancy issues by maintaining state throughout sequential diagnosis while preserving all desirable properties of HS-Tree. Evaluations in a problem domain where HS-Tree is the state-of-the-art diagnostic method reveal stable and significant time savings achieved by DynamicHS.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54973" method="post"> <input type="hidden" name="id" value="54973" /> <div id='downloadlink54973' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54973').click(function () { $('form#downloadform54973').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54974">Explaining Non-Acceptability in Abstract Argumentation</a></div> <label>Authors</label> <div class="value authors">Zeynep G. Saribatur, Johannes P. Wallner, Stefan Woltran</div> <label>Pages</label> <div class="value pages">881 - 888</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200179</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>Argumentation frameworks (AFs) provide a central approach to perform reasoning in many formalisms within argumentation in Artificial Intelligence (AI). Semantics for AFs specify criteria under which sets of arguments can be deemed acceptable, with the notion of admissibility being at the core of main semantics for AFs. A fundamental reasoning task is to find an admissible set containing a queried argument, called credulous acceptance under admissibility. While such a set explains how to argue in favour of a queried argument, finding an explanation in the negative case, i.e., answering why a queried argument is not credulously accepted under admissibility, is less immediate. In this paper, we approach this problem by considering subframeworks of a given AF as witnesses for non-acceptability. Due to the non-monotonicity of semantics for AFs, this requires that every expansion of the witnessing subframework must preserve non-acceptance of the argument—otherwise the subframework would not give sufficient reason for rejection. Among our main contributions (i) we show that this notion of witnessing subframeworks is connected to strong admissibility of AFs, (ii) we investigate the complexity of finding small such subframeworks, and (iii) we extend a recently proposed framework for abstraction in the declarative answer set programming paradigm in order to compute rejecting subframeworks. The resulting system is thus able to deliver explanations also in the case of non-acceptance and we provide a first empirical study that shows the feasibility of our approach.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54974" method="post"> <input type="hidden" name="id" value="54974" /> <div id='downloadlink54974' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54974').click(function () { $('form#downloadform54974').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54975">A Conditional Perspective for Iterated Belief Contraction</a></div> <label>Authors</label> <div class="value authors">Kai Sauerwald, Gabriele Kern-Isberner, Christoph Beierle</div> <label>Pages</label> <div class="value pages">889 - 896</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200180</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>According to Boutillier, Darwiche and Pearl and others, principles for iterated revision can be characterised in terms of changing beliefs about conditionals. For iterated contraction, a similar formulation is not known. In particular, the characterisation for revision does not immediately yield a characterisation for contraction, because in the setting of iterated belief change, revision and contraction are not easily interdefinable. In this article, we develop two axiomatisations of iterated contraction, the first one in terms of changing conditional beliefs, and the second one by employing a new notion of equivalence. We prove that each of these new sets of postulates conforms semantically to the class of operators like the ones given by Konieczny and Pino Pérez for iterated contraction.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54975" method="post"> <input type="hidden" name="id" value="54975" /> <div id='downloadlink54975' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54975').click(function () { $('form#downloadform54975').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54976">Complexity of Possible and Necessary Existence Problems in Abstract Argumentation</a></div> <label>Authors</label> <div class="value authors">Kenneth Skiba, Daniel Neugebauer, J&#246;rg Rothe</div> <label>Pages</label> <div class="value pages">897 - 904</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200181</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>This work focuses on generalizing the existence problems for extensions in abstract argumentation to incomplete argumentation frameworks. In this extended model, incomplete or conflicting knowledge about the state of the arguments and attacks are allowed. We propose possible and necessary variations of the existence and nonemptiness problems, originally defined for (complete) argumentation frameworks, to extend these problems to incomplete argumentation frameworks. While the computational complexity of existence problems is already known for the standard model, we provide a full analysis of the complexity for incomplete argumentation frameworks using the most prominent semantics, namely, the conflict-free, admissible, complete, grounded, preferred, and stable semantics. We show that the complexity rises from NP-completeness to ∏<sup arrange="stack">p</sup><sub arrange="stack">2</sub>-completeness for most “necessary” problem variants when uncertainty is allowed.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54976" method="post"> <input type="hidden" name="id" value="54976" /> <div id='downloadlink54976' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54976').click(function () { $('form#downloadform54976').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54977">How Much Knowledge Is in a Knowledge Base? Introducing Knowledge Measures (Preliminary Report)</a></div> <label>Authors</label> <div class="value authors">Umberto Straccia</div> <label>Pages</label> <div class="value pages">905 - 912</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200182</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>In this work we address the following question: can we measure how much knowledge a knowledge base represents?</p> <p>We answer to this question (i) by describing properties (axioms) that a knowledge measure we believe should have in measuring the amount of knowledge of a knowledge base (kb); and (ii) provide a concrete example of such a measure, based on the notion of entropy.</p> <p>We also introduce related kb notions such as (i) accuracy; (ii) conciseness; and (iii) Pareto optimality. Informally, they address the following questions: (i) how precise is a kb in describing the actual world? (ii) how succinct is a kb w.r.t. the knowledge it represents? and (iii) can we increase accuracy without decreasing conciseness, or vice-versa?</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54977" method="post"> <input type="hidden" name="id" value="54977" /> <div id='downloadlink54977' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54977').click(function () { $('form#downloadform54977').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54978">Minimality and Comparison of Sets of Multi-Attribute Vectors</a></div> <label>Authors</label> <div class="value authors">Federico Toffano, Nic Wilson</div> <label>Pages</label> <div class="value pages">913 - 920</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200183</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>In a decision-making problem, there is often some uncertainty regarding the user preferences. We assume a parameterised utility model, where in each scenario we have a utility function over alternatives, and where each scenario represents a possible user preference model consistent with the input preference information. With a set A of alternatives available to the decision maker, we can consider the associated utility function, expressing, for each scenario, the maximum utility among the alternatives. We consider two main problems: firstly, finding a minimal subset of A that is equivalent to it, i.e., that has the same utility function. We show that for important classes of preference models, the set of so-called possibly strictly optimal alternatives is the unique minimal equivalent subset. Secondly, we consider how to compare A to another set of alternatives B, where A and B correspond to different initial decision choices. We derive mathematical results that allow different computational techniques for these two problems, using linear programming, and especially, with a novel approach using the extreme points of the epigraph of the utility function.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54978" method="post"> <input type="hidden" name="id" value="54978" /> <div id='downloadlink54978' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54978').click(function () { $('form#downloadform54978').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54979">A Hash Learning Framework for Search-Oriented Knowledge Graph Embedding</a></div> <label>Authors</label> <div class="value authors">Meng Wang, Tongtong Wu, Guilin Qi</div> <label>Pages</label> <div class="value pages">921 - 928</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200184</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>Knowledge graph representation learning, also called knowledge graph embedding, is the task of mapping entities and relations into a low-dimensional, continuous vector space, and, as a result, can support various machine learning models to perform knowledge completion tasks with good performance and robustness. However, most of existing embedding models focus on improving the link prediction accuracy while ignoring the time-efficiency in search-intensive applications over large-scale knowledge graphs. To tackle this problem, in this paper, we encode knowledge graph into Hamming space and introduce a novel HAsh Learning Framework (HALF) for search-oriented knowledge graph embedding. The proposed method can be applied to recent various knowledge graph embedding models for accelerating the computation of searching embeddings by utilizing the bitwise operations (XNOR and Bitcount). Experimental results on benchmark datasets demonstrate the effectiveness of our proposed method, which gets a bonus of speed-up in the searching embeddings while the accuracy and scalability of the original model are basically maintained.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54979" method="post"> <input type="hidden" name="id" value="54979" /> <div id='downloadlink54979' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54979').click(function () { $('form#downloadform54979').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54980">Toward Metrics for Differentiating Out-of-Distribution Sets</a></div> <label>Authors</label> <div class="value authors">Mahdieh Abbasi, Changjian Shui, Arezoo Rajabi, Christian Gagn&#233;, Rakesh B. Bobba</div> <label>Pages</label> <div class="value pages">929 - 936</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200185</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>Vanilla CNNs, as uncalibrated classifiers, suffer from classifying out-of-distribution (OOD) samples nearly as confidently as in-distribution samples. To tackle this challenge, some recent works have demonstrated the gains of leveraging available OOD sets for training end-to-end calibrated CNNs. However, a critical question remains unanswered in these works: how to differentiate OOD sets for selecting the most effective one(s) that induce training such CNNs with high detection rates on unseen OOD sets? To address this pivotal question, we provide a criterion based on generalization errors of Augmented-CNN, a vanilla CNN with an added extra class employed for rejection, on in-distribution and unseen OOD sets. However, selecting the most effective OOD set by directly optimizing this criterion incurs a huge computational cost. Instead, we propose three novel computationally-efficient metrics for differentiating between OOD sets according to their “protection” level of in-distribution sub-manifolds. We empirically verify that the most protective OOD sets – selected according to our metrics – lead to A-CNNs with significantly lower generalization errors than the A-CNNs trained on the least protective ones. We also empirically show the effectiveness of a protective OOD set for training well-generalized confidence-calibrated vanilla CNNs. These results confirm that 1) all OOD sets are not equally effective for training well-performing end-to-end models (i.e., A-CNNs and calibrated CNNs) for OOD detection tasks and 2) the protection level of OOD sets is a viable factor for recognizing the most effective one. Finally, across the image classification tasks, we exhibit A-CNN trained on the most protective OOD set can also detect black-box FGS adversarial examples as their distance (measured by our metrics) is becoming larger from the protected sub-manifolds.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54980" method="post"> <input type="hidden" name="id" value="54980" /> <div id='downloadlink54980' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54980').click(function () { $('form#downloadform54980').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54981">Generative Adversarial Networks for Mitigating Biases in Machine Learning Systems</a></div> <label>Authors</label> <div class="value authors">Adel Abusitta, Esma A&#239;meur, Omar Abdel Wahab</div> <label>Pages</label> <div class="value pages">937 - 944</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200186</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>In this paper, we propose a new framework for mitigating biases in machine learning systems. The problem of the existing mitigation approaches is that they are model-oriented in the sense that they focus on tuning the training algorithms to produce fair results, while overlooking the fact that the training data can itself be the main reason for biased outcomes. Technically speaking, two essential limitations can be found in such model-based approaches: 1) the mitigation cannot be achieved without degrading the accuracy of the machine learning models, and 2) when the data used for training are largely biased, the training time automatically increases so as to find suitable learning parameters that help produce fair results. To address these shortcomings, we propose in this work a new framework that can largely mitigate the biases and discriminations in machine learning systems while at the same time enhancing the prediction accuracy of these systems. The proposed framework is based on conditional Generative Adversarial Networks (cGANs), which are used to generate new synthetic fair data with selective properties from the original data. We also propose a framework for analyzing data biases, which is important for understanding the amount and type of data that need to be synthetically sampled and labeled for each population group. Experimental results show that the proposed solution can efficiently mitigate different types of biases, while at the same time enhance the prediction accuracy of the underlying machine learning model.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54981" method="post"> <input type="hidden" name="id" value="54981" /> <div id='downloadlink54981' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54981').click(function () { $('form#downloadform54981').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54982">Understanding and Training Deep Diagonal Circulant Neural Networks</a></div> <label>Authors</label> <div class="value authors">Alexandre Araujo, Benjamin Negrevergne, Yann Chevaleyre, Jamal Atif</div> <label>Pages</label> <div class="value pages">945 - 952</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200187</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>In this paper, we study deep diagonal circulant neural networks, which are deep neural networks in which weight matrices are the product of diagonal and circulant ones. Besides making a theoretical analysis of their expressivity, we introduce principled techniques for training these models: we devise an initialization scheme and propose a smart use of non-linearity functions in order to train deep diagonal circulant networks. Furthermore, we show that these networks outperform recently introduced deep networks with other types of structured layers. We conduct a thorough experimental study to compare the performance of deep diagonal circulant networks with state-of-the-art models based on structured matrices and with dense models. We show that our models achieve better accuracy than other structured approaches while requiring 2x fewer weights than the next best approach. Finally, we train compact and accurate deep diagonal circulant networks on a real world video classification dataset with over 3.8 million training examples.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54982" method="post"> <input type="hidden" name="id" value="54982" /> <div id='downloadlink54982' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54982').click(function () { $('form#downloadform54982').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54983">Ensemble Knowledge Distillation for Learning Improved and Efficient Networks</a></div> <label>Authors</label> <div class="value authors">Umar Asif, Jianbin Tang, Stefan Harrer</div> <label>Pages</label> <div class="value pages">953 - 960</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200188</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>Ensemble models comprising of deep Convolutional Neural Networks (CNN) have shown significant improvements in model generalization but at the cost of large computation and memory requirements. In this paper, we present a framework for learning compact CNN models with improved classification performance and model generalization. For this, we propose a CNN architecture of a compact student model with parallel branches which are trained using ground truth labels and information from high capacity teacher networks in an ensemble learning fashion. Our framework provides two main benefits: i) Distilling knowledge from different teachers into the student network promotes heterogeneity in learning features at different branches of the student network and enables the network to learn diverse solutions to the target problem. ii) Coupling the branches of the student network through ensembling encourages collaboration and improves the quality of the final predictions by reducing variance in the network outputs. Experiments on the well established CIFAR-10 and CIFAR-100 datasets show that our Ensemble Knowledge Distillation (EKD) improves classification accuracy and model generalization especially in situations with limited training data. Experiments also show that our EKD based compact networks outperform in terms of mean accuracy on the test datasets compared to other knowledge distillation based methods.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54983" method="post"> <input type="hidden" name="id" value="54983" /> <div id='downloadlink54983' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54983').click(function () { $('form#downloadform54983').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54984">Compressed k-Nearest Neighbors Ensembles for Evolving Data Streams</a></div> <label>Authors</label> <div class="value authors">Maroua Bahri, Albert Bifet, Silviu Maniu, Rodrigo F. de Mello, Nikolaos Tziortziotis</div> <label>Pages</label> <div class="value pages">961 - 968</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200189</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>The unbounded and multidimensional nature, the evolution of data distributions with time, and the requirement of single-pass algorithms comprise the main challenges of data stream classification, which makes it impossible to infer learning models in the same manner as for batch scenarios. Data dimensionality reduction arises as a key factor to transform and select only the most relevant features from those streams in order to reduce algorithm space and time demands. In that context, Compressed Sensing (CS) encodes an input signal into lower-dimensional space, guaranteeing its reconstruction up to some distortion factor ϵ. This paper employs CS on data streams as a pre-processing step to support a k-Nearest Neighbors (kNN) classification algorithm, one of the most often used algorithms in the data stream mining area – all this while ensuring the key properties of CS hold. Based on topological properties, we show that our classification algorithm also preserves the neighborhood (withing an ϵ factor) of kNN after reducing the stream dimensionality with CS. As a consequence, end-users can set an acceptable error margin while performing such projections for kNN. For further improvements, we incorporate this method into an ensemble classifier, Leveraging Bagging, by combining a set of different CS matrices which increases the diversity inside the ensemble. An extensive set of experiments is performed on various datasets, and the results were compared against those yielded by current state-of-the-art approaches, confirming the good performance of our approaches.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54984" method="post"> <input type="hidden" name="id" value="54984" /> <div id='downloadlink54984' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54984').click(function () { $('form#downloadform54984').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54985">End-To-End Speech Emotion Recognition Based on Time and Frequency Information Using Deep Neural Networks</a></div> <label>Authors</label> <div class="value authors">Ali Bakhshi, Aaron S.W. Wong, Stephan Chalup</div> <label>Pages</label> <div class="value pages">969 - 975</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200190</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>We propose a speech emotion recognition system based on deep neural networks, operating on raw speech data in an end-to-end manner to predict continuous emotions in arousal-valence space. The model is trained using time and frequency information of speech recordings of the publicly available part of the multi-modal RECOLA database. We use the Concordance Correlation Coefficient (CCC) as it was proposed by the Audio-Visual Emotion Challenges to measure the similarity between the network prediction and gold-standard. The CCC prediction results of our model outperform the results achieved by other state-of-the-art end-to-end models. The innovative aspect of our study is an end-to-end approach to using data that previously was mostly used by approaches involving combinations of pre-processing or post-processing. Our study used only a small subset of the RECOLA dataset and obtained better results than previous studies that used the full dataset.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54985" method="post"> <input type="hidden" name="id" value="54985" /> <div id='downloadlink54985' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54985').click(function () { $('form#downloadform54985').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> <div class="bookseriesvolumearticlelistitem"> <div class="content"> <div class="cover"></div> <div class="metadata"> <div class="expandable"> <div class="value title"><a href="/volumearticle/54986">Integrating Network Embedding and Community Outlier Detection via Multiclass Graph Description</a></div> <label>Authors</label> <div class="value authors">Sambaran Bandyopadhyay, Saley Vishal Vivek, M.N. Murty</div> <label>Pages</label> <div class="value pages">976 - 983</div> <label>DOI</label> <div class="value doi">10.3233/FAIA200191</div> <label>Category</label> <div class="value category">Research Article</div> <div class="abstract"> <b>Abstract</b><br /> <section> <p>Network (or graph) embedding is the task to map the nodes of a graph to a lower dimensional vector space, such that it preserves the graph properties and facilitates the downstream network mining tasks. Real world networks often come with (community) outlier nodes, which behave differently from the regular nodes of the community. These outlier nodes can affect the embedding of the regular nodes, if not handled carefully. In this paper, we propose a novel unsupervised graph embedding approach (called DMGD) which integrates outlier and community detection with node embedding. We extend the idea of deep support vector data description to the framework of graph embedding when there are multiple communities present in the given network, and an outlier is characterized relative to its community. We also show the theoretical bounds on the number of outliers detected by DMGD. Our formulation boils down to an interesting minimax game between the outliers, community assignments and the node embedding function. We also propose an efficient algorithm to solve this optimization framework. Experimental results on both synthetic and real world networks show the merit of our approach compared to state-of-the-arts.</p> </section> </div> </div><div class="expandbuttons"><div class="expand">&darr; more</div><div class="collapse">&uarr; less</div></div> </div> <div class="actions"> <form action="/Download/Pdf" id="downloadform54986" method="post"> <input type="hidden" name="id" value="54986" /> <div id='downloadlink54986' class="button getpdf">Download </div> </form> <script type="text/javascript"> $(function () { $('div#downloadlink54986').click(function () { $('form#downloadform54986').submit(); }); }); </script> <div class="button openaccesslicense"> <a rel="license" target="_blank" title="This work is licensed under a Creative Commons License" href="https://creativecommons.org/licenses/by-nc/4.0/deed.en_US"> <img alt="Creative Commons License" style="border-width: 0" src="https://i.creativecommons.org/l/by-nc/4.0/80x15.png" /></a></div> </div> </div> </div> </div> <div class="contentspagination bottom"> <div class="pagination PagedList-pager"><ul><li class="previous PagedList-skipToFirst"><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=1" title="Show first page">&laquo;</a></li><li class="PagedList-skipToPrevious"><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=4" title="Show previous page">&lsaquo;</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=1" title="Show page 1">1</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=2" title="Show page 2">2</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=3" title="Show page 3">3</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=4" title="Show page 4">4</a></li><li class="active"><a>5</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=6" title="Show page 6">6</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=7" title="Show page 7">7</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=8" title="Show page 8">8</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=9" title="Show page 9">9</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=10" title="Show page 10">10</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=11" title="Show page 11">11</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=12" title="Show page 12">12</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=13" title="Show page 13">13</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=14" title="Show page 14">14</a></li><li><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=15" title="Show page 15">15</a></li><li class="disabled PagedList-ellipses"><a>&#8230;</a></li><li class="PagedList-skipToNext"><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=6" title="Show next page">&rsaquo;</a></li><li class="next PagedList-skipToLast"><a data-ajax="true" data-ajax-method="get" data-ajax-mode="replace" data-ajax-update="#unobtrusive" href="/Publication/Descendants/54861?page=17" title="Show last page">&raquo;</a></li></ul></div> </div> <script type="text/javascript"> initExpandBoxesById("volumearticles", 10); </script>