Recommender systems employ recommendation algorithms to predict users’ preferences to items. These preferences are often represented as numerical ratings. However, existing recommender systems seldom ...suggest the appropriate behavior together with the numerical prediction, nor do they consider various types of costs in the recommendation process. In this paper, we propose a regression-based three-way recommender system that aims to minimize the average cost by adjusting the thresholds for different behaviors. This is undertaken using a step-by-step approach, starting with simple problems and progressing to more complex ones. First, we employ memory-based regression approaches for binary recommendation to minimize the loss. Next, we consider misclassification costs and adjust the approaches to minimize the average cost. Finally, we introduce coupon distribution action with promotion cost, and propose two optimal threshold-determination approaches based on the three-way decision model. From the viewpoint of granular computing, a three-way decision is a good tradeoff between the numerical rating and binary recommendation. Experimental results on the well-known MovieLens data set show that threshold settings are critical to the performance of the recommender, and that our approaches can compute unique optimal thresholds.
•We propose a framework integrating three-way decision and random forests.•We introduce a new recommender action to consult the user for the choice.•We build a random forest to predict the ...probability that a user likes an item.•The three-way thresholds are optimal for both the training set and the testing set.
Recommender systems attempt to guide users in decisions related to choosing items based on inferences about their personal opinions. Most existing systems implicitly assume the underlying classification is binary, that is, a candidate item is either recommended or not. Here we propose an alternate framework that integrates three-way decision and random forests to build recommender systems. First, we consider both misclassification cost and teacher cost. The former is paid for wrong recommender behaviors, while the latter is paid to actively consult the user for his or her preferences. With these costs, a three-way decision model is built, and rational settings for positive and negative threshold values α* and β* are computed. We next construct a random forest to compute the probability P that a user will like an item. Finally, α*,0.35em0exβ*, and P are used to determine the recommender’s behavior. The performance of the recommender is evaluated on the basis of an average cost. Experimental results on the well-known MovieLens data set show that the (α*, β*)-pair determined by three-way decision is optimal not only on the training set, but also on the testing set.
•We incorporate sentiment analysis and user reliability into recommendation.•The user reliability adjusts the weights of rating and sentiment information.•Our algorithm outperforms the ...state-of-the-art algorithms on eight Amazon datasets.
Recommender systems aim at predicting users’ preferences based on abundant information, such as user ratings, demographics, and reviews. Although reviews are sparser than ratings, they provide more detailed and reliable information about users’ true preferences. Currently, reviews are often used to improve the explainability of recommender systems. In this paper, we propose the sentiment based matrix factorization with reliability (SBMF+R) algorithm to leverage reviews for prediction. First, we develop a sentiment analysis approach using a new star-based dictionary construction technique to obtain the sentiment score. Second, we design a user reliability measure that combines user consistency and the feedback on reviews. Third, we incorporate the ratings, reviews, and feedback into a probabilistic matrix factorization framework for prediction. Experiments on eight Amazon datasets demonstrated that SBMF+R is more accurate than state-of-the-art algorithms.
This paper proposes a new measure for recommendation through integrating Triangle and Jaccard similarities. The Triangle similarity considers both the length and the angle of rating vectors between ...them, while the Jaccard similarity considers non co-rating users. We compare the new similarity measure with eight state-of-the-art ones on four popular datasets under the leave-one-out scenario. Results show that the new measure outperforms all the counterparts in terms of the mean absolute error and the root mean square error.
Heating is a knotty factor contributing to device degradation of flexible organic solar cells (FOSCs), and thermal regulation plays a crucial role in the realization of long operational lifetime. ...Herein, a passive cooling strategy for stable FOSCs is proposed by boosting the optical‐thermal radiative transfer to reduce the insufficient thermal dissipation and the elevated temperature caused by irradiation‐induced heating, while retaining their flexibility and portability. A spectrally selective coupling structure consisting of subwavelength hemisphere pattern and distributed Bragg reflector is integrated into FOSCs to collectively enhance out‐coupling of infrared radiation and limit near‐infrared absorption‐induced heat generation, leading to a reduced heat power intensity of 292.5 W cm−2 and the decreased working temperature by 9.6 °C under outdoor sunlight irradiation. The D18:Y6:PC71BM‐based FOSCs achieve a power conversion efficiency of over 17% with a prolonged T80 lifetime as long as one year under real outdoor working conditions. These results represent a new opportunity for enhancing the operational stability of FOSCs.
A spectrally selective coupling structure is integrated into flexible organic solar cells to boost the optical‐thermal radiative transfer in infrared region. The optimized device with efficiency over 17% obtains a 9.6 °C decrease in working temperature under outdoor sunlight irradiation, which prolongs T80 lifetime by over 3 times to as long as one year.
In practical scenarios, abnormal network traffic detection often requires analysis of massive, high-dimensional, and unbalanced data. Popular detection methods waste time by processing each data ...stream separately. In this paper, we propose a multi-granularity abnormal network traffic detection algorithm based on multi-instance learning to address this issue. The bag generation technique randomly assembles a predetermined number of data packets into a bag. The bag mapping technique encodes each bag into a new feature space through clustering. The multi-granularity classification technique filters normal data efficiently at the bag granularity before detecting threats at the instance granularity. Experiments were carried out on five datasets in comparison to three state-of-the-art algorithms. Compared with the competing methods, the results show that the average efficiency of this method is increased by more than 10 ~ 20 times, and the accuracy is slightly lower by 0.1 ~ 0.8%.
Demographics are crucial information for recommender systems (RSs). Most existing demographic-based RSs focus on similarity between user profiles. However, they rarely incorporate demographic data to ...describe an item and establish the connection between items and users. In this paper, we propose the concept of the ideal user group (IUG) as a dynamic label for items. This label indicates the users who are most suitable for an item, based on the demographics of its historical customers. Unlike a general label (such as genre or language), the IUG is dynamically changing with the distribution of historical user demographics and is built based on demographic information that undergoes a split-combine process. To validate our method’s effectiveness, we propose an IUG-based neural collaborative filtering (IUG-CF) model. Experimental results on three real-world datasets show that the IUG is an effective approach for improving recommendation performance.
•We construct an ideal user group for each item using explicit rating.•The ideal user group for items dynamically changes with increasing ratings.•The proposed model was validated by comparison with the results of SOTA models.
•We propose a MoEP model to estimate the magic barrier of recommender systems.•The uncertainty parameters are estimated instead of being specified by an expert.•The barriers were validated by ...comparison with the results of SOTA approaches.
In machine learning, the Bayesian error is the lower bound of the prediction error induced by data distribution. In recommender systems, this is also known as the magic barrier (MGBR). MGBR estimation is an important issue because the recommended data frequently contain considerable uncertainties that are difficult to quantify. It is possible to determine the extent to which the recommendation algorithm can be optimized by obtaining the MGBR for a given dataset. MGBR estimation generally requires real user ratings that are not affected by external factors such as human emotions and living environment, which can be extremely difficult or even impossible to gather. Existing theoretical approaches based on simple models, such as Gaussian distributions, have limited estimation capabilities. In this paper, we propose a more sophisticated mixture of exponential power (MoEP) model, which enables adaptive parameter selection for intricate uncertainty. To fit the distribution of the real data, we constructed a flexible learning model that automatically adjusts super- or sub-Gaussian uncertainties using the MoEP components. To select parameters adaptively, we employed an expectation-maximization algorithm to infer the parameters of the components. To estimate the MGBR, we explored an approach for calculating the lower bound of the prediction error under the guidance of a probability model. Experiments on the four datasets validated the rationality of the proposed method. The results show that the MGBR estimated using the new model is marginally lower than the prediction error of state-of-the-art algorithms.
Federated learning is vulnerable to poisoning attacks due to the inability to verify the authenticity of local data. Existing robust federated learning methods maintain a global model by discarding ...potentially risky local updates. However, they generally assume that the server knows the number of potentially abnormal clients. In this paper, we propose a robust federated learning method based on voting and scaling that relaxes such assumption. Malicious updates usually manifest in abnormal direction and magnitude. On one hand, the server computes the relative-angle between the target and other local updates. Angles greater than 90° are considered negative votes, otherwise positive votes. If the negative votes exceed a predefined threshold, the target is considered abnormal. On the other hand, the server computes the magnitude median of the remaining updates after filtering out updates in abnormal directions. The magnitudes of local updates above/below the median are scaled down/increased. Experiments are carried out on five datasets in comparison to five state-of-the-art algorithms. Results on the two metrics of poisoning and main task rates show that our method can effectively improve the robustness of federated learning. Source codes are available at https://github.com/liangxyswpu/lxyCode.
•We propose two mechanisms to improve the robustness of federated learning.•We do not require the central server to collect a clean training dataset.•We do not need to assume that the number of malicious clients is known.
Text classification based on graph neural networks (GNNs) has been widely studied by virtue of its potential to capture complex and across-granularity relations among texts of different types from ...learning on a text graph. Existing methods typically construct text graphs based on words-documents to capture relevant intra-class document representations among the same documents via words-words and words-documents propagation. However, a natural problem is that polysemy words in documents may become an information medium between documents of different categories, promoting heterophily information propagation. The performance of text classification will be somewhat constrained by this issue. This paper proposes a novel text classification method based on GNN from multi-granular topic-aware perspective, referred to as Text-MGNN. Specifically, topic nodes are introduced to build a triple node set of "word, document, topic," and multi-granularity relations are modeled on a text graph for this triple node set. The introduction of topic nodes has three significant advantages. The first is to strengthen the propagation of topics, words, and documents. The second is to enhance class-aware representation learning. The final is to mitigate the effect of heterophily information caused by polysemy words. Extensive experiments are conducted on three real-world datasets. Results validate that our proposed method outperforms 11 baselines methods.