Filters








7,165 Hits in 7.0 sec

Sequential Recommender via Time-aware Attentive Memory Network [article]

Wendi Ji, Keqiang Wang, Xiaoling Wang, TingWei Chen, Alexandra Cristea
2020 arXiv   pre-print
Our approach is scalable for candidate retrieval tasks and can be viewed as a non-linear generalization of latent factorization for dot-product based Top-K recommendation.  ...  We use the proposed time-aware GRU network to learn the short-term intent and maintain prior records in user memory.  ...  Recent research [21] explores the influence of different time intervals on next item prediction and proposes a time-aware self-attentive model.  ... 
arXiv:2005.08598v2 fatcat:zgyhwkpboffsvayn45uluqvbu4

Adaptive User Modeling with Long and Short-Term Preferences for Personalized Recommendation

Zeping Yu, Jianxun Lian, Ahmad Mahmoody, Gongshen Liu, Xing Xie
2019 Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence  
upgraded for better user modeling.  ...  In this paper, we improve the traditional RNN structure by proposing a time-aware controller and a content-aware controller, so that contextual information can be well considered to control the state transition  ...  Acknowledgments We gratefully thank the Microsoft Audience Network (MSAN) team, especially Yajun Wang and Mehul Parsana for their contributions and support for experiments and applications in native advertisement  ... 
doi:10.24963/ijcai.2019/585 dblp:conf/ijcai/YuLML019 fatcat:24k6tw5yhfgonbnnqdvyvcquhi

TLSAN: Time-aware Long- and Short-term Attention Network for Next-item Recommendation

Jianqing Zhang, Dongjing Wang, Dongjin Yu
2021 Neurocomputing  
taste that is related to the "time-aggregation" phenomenon ("personalized time-aggregation"), and 3) users' short-term interests play an important role in the next item prediction/recommendation.  ...  preferences and performing time-sensitive next-item recommendation.  ...  For the next-item recommendation, user preferences are not fixed but change with time.  ... 
doi:10.1016/j.neucom.2021.02.015 fatcat:53ktosol6bgsdk34y7tycr76qu

Histograms of sequences: a novel representation for human interaction recognition

Aytac Cavent, Nazli Ikizler-Cinbis
2018 IET Computer Vision  
Our framework involves extracting visual features from the videos first, and then mining sequences of the visual features that occur consequently in space and time.  ...  This study presents a novel representation based on hierarchical histogram of local feature sequences for human interaction recognition.  ...  Acknowledgments This research was supported in part by TUBITAK Career Development Award 112E149. References  ... 
doi:10.1049/iet-cvi.2017.0471 fatcat:ykkhaemwyfekrdk623aozwff44

Time Interval-enhanced Graph Neural Network for Shared-account Cross-domain Sequential Recommendation [article]

Lei Guo, Jinyu Zhang, Li Tang, Tong Chen, Lei Zhu, Hongzhi Yin
2022 arXiv   pre-print
Moreover, to further enhance item- and account-level representation learning, we incorporate the time interval into the message passing, and design an account-aware self-attention module for learning items  ...  cross-domain graph structure. 3) None existing studies consider the time interval information among items, which is essential in the sequential recommendation for characterizing different items and learning  ...  For evaluation, we treat the last item in each sequence for each domain as the ground truth, and intend to evaluate the ability of our model in predicting the next item given one account's historical behavior  ... 
arXiv:2206.08050v1 fatcat:qxxurwlepjb2pm3fxgqmz6pbma

Parallel Split-Join Networks for Shared-account Cross-domain Sequential Recommendations [article]

Wenchao Sun and Muyang Ma and Pengjie Ren and Yujie Lin and Zhumin Chen and Zhaochun Ren and Jun Ma and Maarten de Rijke
2021 arXiv   pre-print
First, we need to identify the behavior associated with different users and different user roles under the same account in order to recommend the right item to the right user role at the right time.  ...  Sequential recommendation is a task in which one models and uses sequential information about user behavior for recommendation purposes.  ...  items in a sequence have a larger impact on the next item.  ... 
arXiv:1910.02448v4 fatcat:dkmd2mj34bbtjp7fd2tifhan3u

Using Audio-Derived Affective Offset to Enhance TV Recommendation

Sven Ewan Shepstone, Zheng-Hua Tan, Soren Holdt Jensen
2014 IEEE transactions on multimedia  
We show how this affective offset can be used within a framework for providing recommendations for TV programs.  ...  An initial TV content item is then displayed to the user based on the extracted mood profile.  ...  ACKNOWLEDGMENT The authors would like to thank the Swiss Center for Affective Sciences for allowing us to use the GEMEP database.  ... 
doi:10.1109/tmm.2014.2337845 fatcat:iu4e3xukebax5gyqwjz6eukuey

Discrete Event, Continuous Time RNNs [article]

Michael C. Mozer, Denis Kazakov, Robert V. Lindsey
2017 arXiv   pre-print
We investigate recurrent neural network architectures for event-sequence processing.  ...  Our surprising results point both to the robustness of GRU and LSTM architectures for handling continuous time, and to the potency of incorporating continuous dynamics into neural architectures.  ...  We consider standard tasks for event sequences that include classification, prediction of the next event given the time lag from the previous event, and prediction of the time lag to the next event.  ... 
arXiv:1710.04110v1 fatcat:p2i5e7hzlrfx3go4wgjfq2aloe

Geometric Hawkes Processes with Graph Convolutional Recurrent Neural Networks

Jin Shang, Mingxuan Sun
2019 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
Incorporating additional geometric structure in the form of graphs into Hawkes processes is an effective and efficient way for improving model prediction accuracy.  ...  Hawkes processes are popular for modeling correlated temporal sequences that exhibit mutual-excitation properties.  ...  The authors would also like to thank Yichen Wang and Le Song from Georgia Tech for their helpful discussions.  ... 
doi:10.1609/aaai.v33i01.33014878 fatcat:3nbfznmb4ffnbpg2wd5qsp4ey4

On Overfitting and Asymptotic Bias in Batch Reinforcement Learning with Partial Observability (Extended Abstract)

Vincent Francois-Lavet, Guillaume Rabusseau, Joelle Pineau, Damien Ernst, Raphael Fonteneau
2020 Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence  
In the context of reinforcement learning with partial observability, this paper provides an analysis of the tradeoff between these two error sources.  ...  In particular, our theoretical analysis formally characterizes how a smaller state representation increases the asymptotic bias while decreasing the risk of overfitting.  ...  [Hochreiter and Schmidhuber, 1997] with newly introduced time gates to model time intervals between two successive user behaviors for the next-basket recommendations.  ... 
doi:10.24963/ijcai.2020/695 dblp:conf/ijcai/0001Z20 fatcat:yx2wihhuobgmjjh4aevkbr33g4

Fine-Grained Spoiler Detection from Large-Scale Review Corpora

Mengting Wan, Rishabh Misra, Ndapa Nakashole, Julian McAuley
2019 Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics  
This paper presents computational approaches for automatically detecting critical plot twists in reviews of media products.  ...  Third, inspired by these findings, we developed an end-to-end neural network architecture to detect spoiler sentences in review corpora.  ...  We incorporate the item-specificity information in the word embedding layer to enhance word representations based on different item (e.g. book) contexts.  ... 
doi:10.18653/v1/p19-1248 dblp:conf/acl/WanMNM19 fatcat:it4axp3bqrfcpbckmyudtnuu4y

Recognizing multi-user activities using wearable sensors in a smart home

Liang Wang, Tao Gu, Xianping Tao, Hanhua Chen, Jian Lu
2011 Pervasive and Mobile Computing  
In this paper, we investigate the problem of recognizing multiuser activities using wearable sensors in a home setting.  ...  Our experimental results show that we achieve an accuracy of 96.41% with CHMM and an accuracy of 87.93% with FCRF, respectively, for recognizing multi-user activities.  ...  For example, the result of CHMM shows that 24.4% of the observation sequence of reading book/magazine is predicted as watching TV (multi-user), 54.4% of the observation sequence of watching TV (single  ... 
doi:10.1016/j.pmcj.2010.11.008 fatcat:sgkjz434x5fe7h6pvbvdfvd5se

Universal Components of Real-world Diffusion Dynamics based on Point Processes [article]

Minkyoung Kim and Raja Jurdak and Dean Paini
2017 arXiv   pre-print
Bursts in human and natural activities are highly clustered in time, suggesting that these activities are influenced by previous events within the social or natural system.  ...  The proposed universality of diffusion can motivate transdisciplinary research and provide a fundamental framework for diffusion models.  ...  In the next section, we propose a high-level sketch of universal components of the prediction framework for disease diffusion.  ... 
arXiv:1706.06282v2 fatcat:zpr72gg4zjbprl2s4hcdc2mo4q

Step-Enhancement of Memory Retention for User Interest Prediction

Fulian Yin, Pei Su, Sitong Li, Long Ye
2020 IEEE Access  
Numerical experiments using real TV viewing data validate the efficiency of our proposed model and methods, which reduce the average prediction error to 0.3, outperforming the traditional models by around  ...  User behavior modeling and interest prediction are always the key elements in preference analysis, product recommendation and personalized service.  ...  In [21] , Zhu et al. constructed a dynamic user interest model by introducing a time span into the exponent of the function, which divided linear time into time sequences to reflect changes in user interest  ... 
doi:10.1109/access.2020.3002225 fatcat:2at4u44innaldlstimcqii64sa

LSTM Networks for Online Cross-Network Recommendations

Dilruk Perera, Roger Zimmermann
2018 Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence  
Third, time aware LSTM cell gates to capture irregular time intervals between user interactions.  ...  Existing models (1) fail to capture complex non-linear relationships in user interactions, and (2) are designed for offline settings hence, not updated online with incoming interactions to capture the  ...  Figure 6 : 6 Prediction accuracy against the dimensionality of embedding layers. • TimePop : TimePop Calculates the most popular K items for a given time interval and recommends them to all users.  ... 
doi:10.24963/ijcai.2018/532 dblp:conf/ijcai/PereraZ18 fatcat:zrl7q3yzczch7idugcajlakw7q
« Previous Showing results 1 — 15 out of 7,165 results