Ryota HIRAISHI Masatoshi YOSHIKAWA Yang CAO Sumio FUJITA Hidehito GOMI
The significance of individuals' location information has been increasing recently, and the utilization of such data has become indispensable for businesses and society. The possible uses of location information include personalized services (maps, restaurant searches and weather forecast services) and business decisions (deciding where to open a store). However, considering that the data could be exploited, users should add random noise using their terminals before providing location data to collectors. In numerous instances, the level of privacy protection a user requires depends on their location. Therefore, in our framework, we assume that users can specify different privacy protection requirements for each location utilizing the adversarial error (AE), and the system computes a mechanism to satisfy these requirements. To guarantee some utility for data analysis, the maximum error in outputting the location should also be output. In most privacy frameworks, the mechanism for adding random noise is public; however, in this problem setting, the privacy protection requirements and the mechanism must be confidential because this information includes sensitive information. We propose two mechanisms to address privacy personalization. The first mechanism is the individual exponential mechanism, which uses the exponential mechanism in the differential privacy framework. However, in the individual exponential mechanism, the maximum error for each output can be used to narrow down candidates of the actual location by observing outputs from the same location multiple times. The second mechanism improves on this deficiency and is called the donut mechanism, which uniformly outputs a random location near the location where the distance from the user's actual location is at the user-specified AE distance. Considering the potential attacks against the idea of donut mechanism that utilize the maximum error, we extended the mechanism to counter these attacks. We compare these two mechanisms by experiments using maps constructed from artificial and real world data.
Yang LIU Yuqi XIA Haoqin SUN Xiaolei MENG Jianxiong BAI Wenbo GUAN Zhen ZHAO Yongwei LI
Speech emotion recognition (SER) has been a complex and difficult task for a long time due to emotional complexity. In this paper, we propose a multitask deep learning approach based on cascaded attention network and self-adaption loss for SER. First, non-personalized features are extracted to represent the process of emotion change while reducing external variables' influence. Second, to highlight salient speech emotion features, a cascade attention network is proposed, where spatial temporal attention can effectively locate the regions of speech that express emotion, while self-attention reduces the dependence on external information. Finally, the influence brought by the differences in gender and human perception of external information is alleviated by using a multitask learning strategy, where a self-adaption loss is introduced to determine the weights of different tasks dynamically. Experimental results on IEMOCAP dataset demonstrate that our method gains an absolute improvement of 1.97% and 0.91% over state-of-the-art strategies in terms of weighted accuracy (WA) and unweighted accuracy (UA), respectively.
With the spread of the broadband Internet and high-performance devices, various services have become available anytime, anywhere. As a result, attention is focused on the service quality and Quality of Experience (QoE) is emphasized as an evaluation index from the user's viewpoint. Since QoE is a subjective evaluation metric and deeply involved with user perception and expectation, quantitative and comparative research was difficult because the QoE study is still in its infancy. At present, after tremendous devoted efforts have contributed to this research area, a shape of the QoE management architecture has become clear. Moreover, not only for research but also for business, video streaming services are expected as a promising Internet service incorporating QoE. This paper reviews the present state of QoE studies with the above background and describes the future prospect of QoE. Firstly, the historical aspects of QoE is reviewed starting with QoS (Quality of Service). Secondly, a QoE management architecture is proposed in this paper, which consists of QoE measurement, QoE assessment, QoS-QoE mapping, QoE modeling, and QoE adaptation. Thirdly, QoE studies related with video streaming services are introduced, and finally individual QoE and physiology-based QoE measurement methodologies are explained as future prospect in the field of QoE studies.
Yuxuan ZHU Yong PENG Yang SONG Kenji OZAWA Wanzeng KONG
In this study we propose a method to perform personal identification (PI) based on Electroencephalogram (EEG) signals, where the used network is named residual and multiscale spatio-temporal convolution neural network (RAMST-CNN). Combined with some popular techniques in deep learning, including residual learning (RL), multi-scale grouping convolution (MGC), global average pooling (GAP) and batch normalization (BN), RAMST-CNN has powerful spatio-temporal feature extraction ability as it achieves task-independence that avoids the complexity of selecting and extracting features manually. Experiments were carried out on multiple datasets, the results of which were compared with methods from other studies. The results show that the proposed method has a higher recognition accuracy even though the network it is based on is lightweight.
Siyang YU Kazuaki KONDO Yuichi NAKAMURA Takayuki NAKAJIMA Masatake DANTSUJI
This article introduces our investigation on learning state estimation in e-learning on the condition that visual observation and recording of a learner's behaviors is possible. In this research, we examined methods of adaptation for a new learner for whom a small number of ground truth data can be obtained.
Mingu KIM Seungwoo HONG Il Hong SUH
Personalized trip planning is a challenging problem given that places of interest should be selected according to user preferences and sequentially arranged while satisfying various constraints. In this study, we aimed to model various uncertain aspects that should be considered during trip planning and efficiently generate personalized plans that maximize user satisfaction based on preferences and constraints. Specifically, we propose a probabilistic itinerary evaluation model based on a hybrid temporal Bayesian network that determines suitable itineraries considering preferences, constraints, and uncertain environmental variables. The model retrieves the sum of time-weighted user satisfaction, and ant colony optimization generates the trip plan that maximizes the objective function. First, the optimization algorithm generates candidate itineraries and evaluates them using the proposed model. Then, we improve candidate itineraries based on the evaluation results of previous itineraries. To validate the proposed trip planning approach, we conducted an extensive user study by asking participants to choose their preferred trip plans from options created by a human planner and our approach. The results show that our approach provides human-like trip plans, as participants selected our generated plans in 57% of the pairs. We also evaluated the efficiency of the employed ant colony optimization algorithm for trip planning by performance comparisons with other optimization methods.
Yasuhiro FUJIWARA Makoto NAKATSUJI Hiroaki SHIOKAWA Takeshi MISHIMA Makoto ONIZUKA
Personalized PageRank (PPR) is a typical similarity metric between nodes in a graph, and node searches based on PPR are widely used. In many applications, graphs change dynamically, and in such cases, it is desirable to perform ad hoc searches based on PPR. An ad hoc search involves performing searches by varying the search parameters or graphs. However, as the size of a graph increases, the computation cost of performing an ad hoc search can become excessive. In this paper, we propose a method called Castanet that offers fast ad hoc searches of PPR. The proposed method features (1) iterative estimation of the upper and lower bounds of PPR scores, and (2) dynamic pruning of nodes that are not needed to obtain a search result. Experiments confirm that the proposed method does offer faster ad hoc PPR searches than existing methods.
Xibin WANG Fengji LUO Chunyan SANG Jun ZENG Sachio HIROKAWA
With the rapid development of information and Web technologies, people are facing ‘information overload’ in their daily lives. The personalized recommendation system (PRS) is an effective tool to assist users extract meaningful information from the big data. Collaborative filtering (CF) is one of the most widely used personalized recommendation techniques to recommend the personalized products for users. However, the conventional CF technique has some limitations, such as the low accuracy of of similarity calculation, cold start problem, etc. In this paper, a PRS model based on the Support Vector Machine (SVM) is proposed. The proposed model not only considers the items' content information, but also the users' demographic and behavior information to fully capture the users' interests and preferences. An improved Particle Swarm Optimization (PSO) algorithm is also proposed to improve the performance of the model. The efficiency of the proposed method is verified by multiple benchmark datasets.
Siyang YU Kazuaki KONDO Yuichi NAKAMURA Takayuki NAKAJIMA Masatake DANTSUJI
Self-paced e-learning provides much more freedom in time and locale than traditional education as well as diversity of learning contents and learning media and tools. However, its limitations must not be ignored. Lack of information on learners' states is a serious issue that can lead to severe problems, such as low learning efficiency, motivation loss, and even dropping out of e-learning. We have designed a novel e-learning support system that can visually observe learners' non-verbal behaviors and estimate their learning states and that can be easily integrated into practical e-learning environments. Three pairs of internal states closely related to learning performance, concentration-distraction, difficulty-ease, and interest-boredom, were selected as targets of recognition. In addition, we investigated the practical problem of estimating the learning states of a new learner whose characteristics are not known in advance. Experimental results show the potential of our system.
Kenta SERIZAWA Sayaka KAMEI Syuhei HAYASHI Satoshi FUJITA
In this paper, a new scheme for personalized web page recommendation using multi-user search engine query information is proposed. Our contribution is a scheme that improves the accuracy of personalization for various types of contents (e.g., documents, images and music) without increasing user burden. The proposed scheme combines “preference footprints” for browsed pages with collaborative filtering. We acquire user interest using words that are relevant to queries submitted by users, attach all user interests to a page as a footprint when it is browsed, and evaluate the relevance of web pages in relation to words in footprints. The performance of the scheme is evaluated experimentally. The results indicate that the proposed scheme improves the precision and recall of previous schemes by 1%-24% and 80%-107%, respectively.
Worapol TANGKOKIATTIKUL Aphirak JANSANG Anan PHONPHOEM
Personal Wi-Fi Hotspot, the Wi-Fi tethering function, is widely deployed on mobile devices to allow other wireless clients to share Internet access via a broadband connection. Its advantages include no connection fee and support of non-3G/LTE devices. However, utilizing this function can rapidly deplete the battery power of the tethering device because both interface connections (3G/LTE and Wi-Fi) are always on. To address this problem, this paper proposes the Energy Management Mechanism for Wi-Fi Tethering Mode on Mobile Devices (EMWT). The mechanism is designed to effectively manage both interfaces by adjusting certain sleep durations according to the incoming traffic. Short, Long, and Deep sleep durations are introduced for saving energy. EMWT can also guarantee the packet delay bound by limiting the maximum sleep period. Five traffic rates, composed of very low, low, medium, high, and very high, are evaluated. NS-3 simulation results reveal that energy savings of up to 52.52% can be achieved with only a slight impact on system performance, in terms of end-to-end delay, throughput, and packet loss.
Connection Service Providers (CSP) are wishing to increase their Return on Investment (ROI) by utilizing the data assets generated by tracking subscriber behaviors. This results in the ability to apply personalized policies, monitor and control the service traffic to subscribers and gain more revenue through the usage of subscriber data with ad networks. In this paper, a system is proposed to monitor and analyze the Internet access of the subscribers of a regional SP in order to classify the subscribers into interest categories from the Interactive Advertising Bureau (IAB) categories. The study employs the categorization engine to build category vectors for all individuals using Internet services through the subscription. The proposal makes it easy to detect changes in the interests of individuals/subscribers over time.
Huifeng GUO Dianhui CHU Yunming YE Xutao LI Xixian FAN
Ranking as an important task in information systems has many applications, such as document/webpage retrieval, collaborative filtering and advertising. The last decade has witnessed a growing interest in the study of learning to rank as a means to leverage training information in a system. In this paper, we propose a new learning to rank method, i.e. BLM-Rank, which uses a linear function to score samples and models the pairwise preference of samples relying on their scores under a Bayesian framework. A stochastic gradient approach is adopted to maximize the posterior probability in BLM-Rank. For industrial practice, we have also implemented the proposed algorithm on Graphic Processing Unit (GPU). Experimental results on LETOR have demonstrated that the proposed BLM-Rank method outperforms the state-of-the-art methods, including RankSVM-Struct, RankBoost, AdaRank-NDCG, AdaRank-MAP and ListNet. Moreover, the results have shown that the GPU implementation of the BLM-Rank method is ten-to-eleven times faster than its CPU counterpart in the training phase, and one-to-four times faster in the testing phase.
Infrastructures for the evaluation of the state of health of individuals using a standardized communication network consisting of advanced instruments and subsequent data analysis have been developed. Here we report that this developed infrastructure has been tested in the field in 100 houses and involving almost 300 users. The communication protocol part of this infrastructure has been standardized as IEEE 11073-20601. Continua Health Alliance, an international not-for-profit industry organization which has nearly 230 member companies, has adopted this IEEE 11073-20601 to establish an ecosystem of interoperable personal connected health systems that empower individuals and organizations to better manage their health and wellness. Currently nearly 100 Continua certified products are available in public including smartphone.
Recent emerging mobile and wearable technologies make it easy to collect personal spatiotemporal data such as activity trajectories in daily life. Publishing real-time statistics over trajectory streams produced by crowds of people is expected to be valuable for both academia and business, answering questions such as “How many people are in Kyoto Station now?” However, analyzing these raw data will entail risks of compromising individual privacy. ε-Differential Privacy has emerged as a well-known standard for private statistics publishing because of its guarantee of being rigorous and mathematically provable. However, since user trajectories will be generated infinitely, it is difficult to protect every trajectory under ε-differential privacy. On the other hand, in real life, not all users require the same level of privacy. To this end, we propose a flexible privacy model of l-trajectory privacy to ensure every desired length of trajectory under protection of ε-differential privacy. We also design an algorithmic framework to publish l-trajectory private data in real time. Experiments using four real-life datasets show that our proposed algorithms are effective and efficient.
SooHyung KIM Daeseon CHOI Seung-Hun JIN Hyunsoo YOON JinWoo SON MyungKeun YOON
New payment technologies are coming that will raise user convenience. To support automatic hands-free payment, merchant devices will collect customer's information from the cloud of payment service providers or customer's smart phones, which should be removed after the transaction. Using Jaccard containment, we propose a proactive security approach of cleaning personal data at merchant-side point-of-sale terminals. We also propose a sampling method to reduce communication overhead by several orders of magnitude.
Rajashree S. SOKASANE Kyungbaek KIM
In these days, recognizing a user personality is an important issue in order to support various personalized services. Besides the conventional phone usage such as call logs, SMS logs and application usages, smart phones can gather the behavior of users by polling various embedded sensors such as GPS sensors. In this paper, we focus on how to predict user attitude based on GPS log data by applying location clustering techniques and extracting features from the location clusters. Through the evaluation with one month-long GPS log data, it is observed that the location-based features, such as number of clusters and coverage of clusters, are correlated with user attitude to some extent. Especially, when SVM is used as a classifier for predicting the dichotomy of user attitudes of MBTI, over 90% F-measure is achieved.
Chen CHEN Chunyan HOU Peng NIE Xiaojie YUAN
Recommendation systems have been widely used in E-commerce sites, social media and etc. An important recommendation task is to predict items that a user will perform actions on with users' historical data, which is called top-K recommendation. Recently, there is huge amount of emerging items which are divided into a variety of categories and researchers have argued or suggested that top-K recommendation of item category could be very beneficial for users to make better and faster decisions. However, the traditional methods encounter some common but crucial problems in this scenario because additional information, such as time, is ignored. The ranking algorithm on graphs and the increasingly growing amount of online user behaviors shed some light on these problems. We propose a construction method of time-aware graphs to use ranking algorithm for personalized recommendation of item category. Experimental results on real-world datasets demonstrate the advantages of our proposed method over competitive baseline algorithms.
Daeseon CHOI Younho LEE Yongsu PARK Seokhyun KIM
People expose their personal information on social network services (SNSs). This paper warns of the dangers of this practice by way of an example. We show that the residence registration numbers (RRNs) of many Koreans, which are very important and confidential personal information analogous to social security numbers in the United States, can be estimated solely from the information that they have made open to the public. In our study, we utilized machine learning algorithms to infer information that was then used to extract a part of the RRNs. Consequently, we were able to extract 45.5% of SNS users' RRNs using a machine learning algorithm and brute-force search that did not consume exorbitant amounts of resources.
Qing DU Yu LIU Dongping HUANG Haoran XIE Yi CAI Huaqing MIN
With the development of the Internet, there are more and more shared resources on the Web. Personalized search becomes increasingly important as users demand higher retrieval quality. Personalized search needs to take users' personalized profiles and information needs into consideration. Collaborative tagging (also known as folksonomy) systems allow users to annotate resources with their own tags (features) and thus provide a powerful way for organizing, retrieving and sharing different types of social resources. To capture and understand user preferences, a user is typically modeled as a vector of tag: value pairs (i.e., a tag-based user profile) in collaborative tagging systems. In such a tag-based user profile, a user's preference degree on a group of tags (i.e., a combination of several tags) mainly depends on the preference degree on every individual tag in the group. However, the preference degree on a combination of tags (a tag-group) cannot simply be obtained from linearly combining the preference on each tag. The combination of a user's two favorite tags may not be favorite for the user. In this article, we examine the limitations of previous tag-based personalized search. To overcome their problems, we model a user profile based on combinations of tags (tag-groups) and then apply it to the personalized search. By comparing it with the state-of-the-art methods, experimental results on a real data set shows the effectiveness of our proposed user profile method.