We propose a novel approach termed ''neighbors for one target (N41). Code is available at https://github.com/wshw4ng/N41, '' which improves anomaly detection accuracy for multivariate time-series data. N41 redefines both the output and input of any detection model by ''single target feature'' and ''its neighboring features,'' respectively. Using a single target feature improves accuracy by allowing the model to focus solely on minimizing the training error (loss function) for that specific feature. In contrast, defining multiple features as outputs distributes the training error across them, potentially leading to less effective error reduction for some features. Incorporating neighboring features with patterns similar to the target feature further enhances accuracy by including relevant features. Evaluation on the real-world HAI22.04 dataset demonstrated that N41 achieved F1 scores 1.7 to 7.5 times higher than those of existing methods.

We proposetime-series aware precision andrecall, which are appropriate for evaluating anomaly detection methods in time-series data. In time-series data, an anomaly corresponds toa series of instances. The conventional metrics, however, overlook this characteristic, so they suffer from a problem of giving a high score to the method that only detects a long anomaly. To overcome the problem, our metrics consider thevariety of the detected anomalies to be more important through two scoring strategies,detection scoring (\ie, how many anomalies are detected) andportion scoring (\ie, how precisely each anomaly is detected). Moreover, our metrics concernambiguous instances, which indicate the instances labeled as 'normal' although they are affected by their precedent anomaly. Our metrics give smaller scores to those instances as they are likely to be anomalous. We demonstrate that our metrics are more suitable for time-series data compared to existing metrics by evaluations using a real-world dataset as well as several examples.\footnoteOur code and the detection results are available at: https://github.com/saurf4ng/TaPR.

We develop a novel framework, named as$l$-injection, to address the sparsity problem of recommender systems. By carefully injecting low values to a selected set of unrated user-item pairs in a user-item matrix, we demonstrate that top-Nrecommendation accuracies of various collaborative filtering (CF) techniques can be significantly and consistently improved. We first adopt the notion ofpre-use preferencesof users toward a vast amount ofunrateditems. Using this notion, we identifyuninterestingitems that have not been rated yet but are likely to receive low ratings from users, and selectively impute them as low values. As our proposed approach is method-agnostic, it can be easily applied to a variety of CF algorithms. Through comprehensive experiments with three real-life datasets (e.g., Movielens, Ciao, and Watcha), we demonstrate that our solution consistently and universally enhances the accuracies of existing CF algorithms (e.g., item-based CF, SVD-based CF, and SVD++) by 2.5 to 5 times on average. Furthermore, our solution improves the running time of those CF methods by 1.2 to 2.3 times when its setting produces the best accuracy. The datasets and codes that we used in the experiments are available at:https://goo.gl/KUrmip.

We study how to improve the accuracy and running time of top-N recommendation with collaborative filtering (CF). Unlike existing works that use mostly rated items (which is only a small fraction in a rating matrix), we propose the notion of pre-use preferences of users toward a vast amount of unrated items. Using this novel notion, we effectively identify uninteresting items that were not rated yet but are likely to receive very low ratings from users, and impute them as zero. This simple-yet-novel zero-injection method applied to a set of carefully-chosen uninteresting items not only addresses the sparsity problem by enriching a rating matrix but also completely prevents uninteresting items from being recommended as top-N items, thereby improving accuracy greatly. As our proposed idea is method-agnostic, it can be easily applied to a wide variety of popular CF methods. Through comprehensive experiments using the Movielens dataset and MyMediaLite implementation, we successfully demonstrate that our solution consistently and universally improves the accuracies of popular CF methods (e.g., item-based CF, SVD-based CF, and SVD++) by two to five orders of magnitude on average. Furthermore, our approach reduces the running time of those CF methods by 1.2 to 2.3 times when its setting produces the best accuracy. The datasets and codes that we used in experiments are available at: https://goo.gl/KUrmip.

Single molecule time trajectories of biomolecules provide glimpses into complex folding landscapes that are difficult to visualize using conventional ensemble measurements. Recent experiments and theoretical analyses have highlighted dynamic disorder in certain classes of biomolecules, whose dynamic pattern of conformational transitions is affected by slower transition dynamics of internal state hidden in a low dimensional projection. A systematic means to analyze such data is, however, currently not well developed. Here we report a new algorithm-Variational Bayes-double chain Markov model (VB-DCMM)-to analyze single molecule time trajectories that display dynamic disorder. The proposed analysis employing VB-DCMM allows us to detect the presence of dynamic disorder, if any, in each trajectory, identify the number of internal states, and estimate transition rates between the internal states as well as the rates of conformational transition within each internal state. Applying VB-DCMM algorithm to single molecule FRET data of H-DNA in 100 mM-Na+ solution, followed by data clustering, we show that at least 6 kinetic paths linking 4 distinct internal states are required to correctly interpret the duplex-triplex transitions of H-DNA.

In a trust network, two users who are connected by a trust relationship tend to have similar interests. Based on this observation, existing trust-aware recommendation methods predict ratings for a target user on unseen items by referencing to ratings of those users who are reachable from the target user in the forward direction of trustor-trustee relationship through the trust network. However, these methods have overlooked the possibility of utilizing the ratings of those users reachable in the backward direction, which may also have similar interests. In this paper, we investigate this possibility by identifying and adding these users to the existing methods when predicting ratings for the target user. We perform a series of experiments and observe that our approach improves the coverage while preserving the accuracy.

A variety of recommendation methods have been proposed to satisfy the performance and accuracy; however, it is fairly difficult to satisfy both of them because there is a trade-off between them. In this paper, we introduce the notion of category experts and propose the recommendation method by exploiting the ratings of category experts instead of those of the users similar to a target user. We also extend the method that uses both the category preference of a target user and his/her similarity to category experts. We show that our method significantly outperforms the existing methods in terms of performance and accuracy through extensive experiments with real-world data.

Recently, product review sitesthat contain user ratings and trust statements such as Epinions.com have become popular, and many attempts have been made to analyze user behavior on those sites. In this paper, we first define three different user groups for each user, i.e. Trustee, Trust or, and Irrelevance which describe different kinds of trust statements. Then we investigate the relationship between each user and users in his/her different user groups by comparing how their ratings are similar on Epinions data sets. Most existing works simply ignored Trust or group when considering a certain user. However, our observations suggest that Trust or group still meaningful that should not be simply disregarded.

In this paper, we propose a new paper ranking algorithm that gives a high rank to papers which is credited by other authoritative papers or published in premier conferences or journals. Also, the proposed algorithm solves a problem that recent papers are rated poorly due to few citations.