Existing approximate nearest neighbor search systems suffer from two fundamental problems that are of practical importance but have not received sufficient attention from the research community. First, although existing systems perform well for the whole database, it is difficult to run a search over a subset of the database. Second, there has been no discussion concerning the performance decrement after many items have been newly added to a system. We develop a reconfigurable inverted index (Rii) to resolve these two issues. Based on the standard IVFADC system, we design a data layout such that items are stored linearly. This enables us to efficiently run a subset search by switching the search method to a linear PQ scan if the size of a subset is small. Owing to the linear layout, the data structure can be dynamically adjusted after new items are added, maintaining the fast speed of the system. Extensive comparisons show that Rii achieves a comparable performance with state-of-the art systems such as Faiss.
This paper presents an empirical exploration of the use of capsule networks for text classification. While it has been shown that capsule networks are effective for image classification, their validity in the domain of text has not been explored. In this paper, we show that capsule networks indeed have the potential for text classification and that they have several advantages over convolutional neural networks. We further suggest a simple routing method that effectively reduces the computational complexity of dynamic routing. We utilized seven benchmark datasets to demonstrate that capsule networks, along with the proposed routing method provide comparable results.
The time series classification literature has expanded rapidly over the last decade, with many new classification approaches published each year. The research focus has mostly been on improving the accuracy and efficiency of classifiers, while their interpretability has been somewhat neglected. Classifier interpretability has become a critical constraint for many application domains and the introduction of the ‘right to explanation’ GDPR EU legislation in May 2018 is likely to further emphasize the importance of explainable learning algorithms. In this work we analyse the state-of-the-art for time series classification, and propose new algorithms that aim to maintain the classifier accuracy and efficiency, but keep interpretability as a key design constraint. We present new time series classification algorithms that advance the state-of-the-art by implementing the following three key ideas: (1) Multiple resolutions of symbolic approximations: we combine symbolic representations obtained using different parameters; (2) Multiple domain representations: we combine symbolic approximations in time (e.g., SAX) and frequency (e.g., SFA) domains; (3) Efficient navigation of a huge symbolic-words space: we adapt a symbolic sequence classifier named SEQL, to make it work with multiple domain representations (e.g., SAX-SEQL, SFA-SEQL), and use its greedy feature selection strategy to effectively filter the best features for each representation. We show that a multi-resolution multi-domain linear classifier, SAX-SFA-SEQL, achieves a similar accuracy to the state-of-the-art COTE ensemble, and to a recent deep learning method (FCN), but uses a fraction of the time required by either COTE or FCN. We discuss the accuracy, efficiency and interpretability of our proposed algorithms. To further analyse the interpretability aspect of our classifiers, we present a case study on an ecology benchmark.
Big data is a term used for a very large data sets that have many difficulties in storing and processing the data. Analysis this much amount of data will lead to information loss. The main goal of this paper is to share data in a way that privacy is preserved while information loss is kept at least. Data that include Government agencies, University details and Medical history etc., are very necessary for an organization to do analysis and predict trends and patterns, but it may prevent the data owner from sharing the data because of privacy regulations [1]. By doing an analysis of several algorithms of Anonymization such as k-anonymity, l-diversity and tcloseness, one can achieve privacy at minimum loss. Admitting these techniques has some limitations. We need to maintain trade-off between privacy and information loss. We introduce a novel approach called Differential Privacy.
A time-domain test for the assumption of second order stationarity of a functional time series is proposed. The test is based on combining individual cumulative sum tests which are designed to be sensitive to changes in the mean, variance and autocovariance operators, respectively. The combination of their dependent $p$-values relies on a joint dependent block multiplier bootstrap of the individual test statistics. Conditions under which the proposed combined testing procedure is asymptotically valid under stationarity are provided. A procedure is proposed to automatically choose the block length parameter needed for the construction of the bootstrap. The finite-sample behavior of the proposed test is investigated in Monte Carlo experiments and an illustration on a real data set is provided.
Certain upper triangular matrices, termed as Parikh matrices, are often used in the combinatorial study of words. Given a word, the Parikh matrix of that word elegantly computes the number of occurrences of certain predefined subwords in that word. In this paper, we compute the Parikh matrix of any word raised to an arbitrary power. Furthermore, we propose canonical decompositions of both Parikh matrices and words into normal forms. Finally, given a Parikh matrix, the relation between its normal form and the normal forms of words in the corresponding M-equivalence class is established.
In this paper, we introduce an embedding model, named CapsE, exploring a capsule network to model relationship triples \textit{(subject, relation, object)}. Our CapsE represents each triple as a 3-column matrix where each column vector represents the embedding of an element in the triple. This 3-column matrix is then fed to a convolution layer where multiple filters are operated to generate different feature maps. These feature maps are used to construct capsules in the first capsule layer. Capsule layers are connected via dynamic routing mechanism. The last capsule layer consists of only one capsule to produce a vector output. The length of this vector output is used to measure the plausibility of the triple. Our proposed CapsE obtains state-of-the-art link prediction results for knowledge graph completion on two benchmark datasets: WN18RR and FB15k-237, and outperforms strong search personalization baselines on SEARCH17 dataset.
The most approaches to Knowledge Base Question Answering are based on semantic parsing. In this paper, we address the problem of learning vector representations for complex semantic parses that consist of multiple entities and relations. Previous work largely focused on selecting the correct semantic relations for a question and disregarded the structure of the semantic parse: the connections between entities and the directions of the relations. We propose to use Gated Graph Neural Networks to encode the graph structure of the semantic parse. We show on two data sets that the graph networks outperform all baseline models that do not explicitly model the structure. The error analysis confirms that our approach can successfully process complex semantic parses.
PatternAttribution is a recent method, introduced in the vision domain, that explains classifications of deep neural networks. We demonstrate that it also generates meaningful interpretations in the language domain.
The concept of Probability of Causation (PC) is critically important in legal contexts and can help in many other domains. While it has been around since 1986, current operationalizations can obtain only the minimum and maximum values of PC, and do not apply for purely observational data. We present a theoretical framework to estimate the distribution of PC from experimental and from purely observational data. We illustrate additional problems of the existing operationalizations and show how our method can be used to address them. We also provide two illustrative examples of how our method is used and how factors like sample size or rarity of events can influence the distribution of PC. We hope this will make the concept of PC more widely usable in practice.