Filters








273 Hits in 2.8 sec

Controllable Recommenders using Deep Generative Models and Disentanglement [article]

Samarth Bhargav, Evangelos Kanoulas
2021 arXiv   pre-print
We show that by updating the disentangled latent space based on user feedback, and by exploiting the generative nature of the recommender, controlled and personalized recommendations can be produced.  ...  We propose using a (semi-)supervised disentanglement objective for this purpose, as well as multiple metrics to evaluate the controllability and the degree of personalization of controlled recommendations  ...  ACKNOWLEDGMENTS The authors would like to thank Mohammad Aliannejadi, Wilker Aziz, Maurits Bleeker, Jin Huang, Antonis Krasakis, Anna Sepliarskaia, Georgios Sidiropoulos and Svitlana Vakulenko for helpful  ... 
arXiv:2110.05056v1 fatcat:eepjuxveebffdb2i7vglqx65ra

Learning Disentangled Representations for Recommendation [article]

Jianxin Ma, Chang Zhou, Peng Cui, Hongxia Yang, Wenwu Zhu
2019 arXiv   pre-print
We further demonstrate that the learned representations are interpretable and controllable, which can potentially lead to a new paradigm for recommendation where users are given fine-grained control over  ...  In this paper, we present the MACRo-mIcro Disentangled Variational Auto-Encoder (MacridVAE) for learning disentangled representations from user behavior.  ...  User-Controllable Recommendation The controllability enabled by the disentangled representations can bring a new paradigm for recommendation.  ... 
arXiv:1910.14238v1 fatcat:paxki7jfdfdrhfngqrglqsx4mm

Controllable Gradient Item Retrieval [article]

Haonan Wang, Chang Zhou, Carl Yang, Hongxia Yang, Jingrui He
2021 arXiv   pre-print
To deal with this problem, we propose a weakly-supervised method that can learn a disentangled item representation from user-item interaction data and ground the semantic meaning of attributes to dimensions  ...  For example, after a customer saw a white dress, she/he wants to buy a similar one but more floral on it.  ...  The left part is designed based on Variational Autoencoder framework [17] , which learns a disentangled item representation from user activities.  ... 
arXiv:2106.00062v1 fatcat:rcg4tkzsefhjpepyaracarzxpe

Intent Disentanglement and Feature Self-supervision for Novel Recommendation [article]

Tieyun Qian, Yile Liang, Qing Li, Xuan Ma, Ke Sun, Zhiyong Peng
2021 arXiv   pre-print
One key property in recommender systems is the long-tail distribution in user-item interactions where most items only have few user feedback.  ...  We then present a unified end-to-end framework to simultaneously optimize accuracy and novelty targets based on the disentangled intent of popularity and that of preference.  ...  training strategy for point-wise and pair-wise objectives based on the disentangled user and item representations. (1) For the point-wise learning, the model takes a triplet (u, v, y uv ) as input.  ... 
arXiv:2106.14388v1 fatcat:m4i73l45fzgzfhp7k3ud4aqc5y

Privacy-preserving Voice Analysis via Disentangled Representations [article]

Ranya Aloufi, Hamed Haddadi, David Boyle
2020 arXiv   pre-print
We leverage disentangled representation learning to explicitly learn independent factors in the raw data.  ...  Despite offering an always-on convenient user experience, VUIs raise new security and privacy concerns for their users.  ...  Disentanglement Models Most prior works on disentangled representation learning are based on well-established frameworks, such as variational autoencoders (VAEs) [40] and generative adversarial models  ... 
arXiv:2007.15064v1 fatcat:oancnsvxlja4zdz2whxef5s3tm

Structured Neural Topic Models for Reviews [article]

Babak Esmaeili, Hongyi Huang, Byron C. Wallace, Jan-Willem van de Meent
2019 arXiv   pre-print
We present Variational Aspect-based Latent Topic Allocation (VALTA), a family of autoencoding topic models that learn aspect-based representations of reviews.  ...  VALTA defines a user-item encoder that maps bag-of-words vectors for combined reviews associated with each paired user and item onto structured embeddings, which in turn define per-aspect topic weights  ...  In NLP, work on learning disentangled representations has been more limited [14] [15] [16] [17] .  ... 
arXiv:1812.05035v2 fatcat:ug22hpmcrbfqxdzys2vq7h7r6q

Toward Controlled Generation of Text [article]

Zhiting Hu, Zichao Yang, Xiaodan Liang, Ruslan Salakhutdinov, Eric P. Xing
2018 arXiv   pre-print
This paper aims at generating plausible natural language sentences, whose attributes are dynamically controlled by learning disentangled latent representations with designated semantics.  ...  With differentiable approximation to discrete text samples, explicit constraints on independent attribute controls, and efficient collaborative learning of generator and discriminators, our model learns  ...  Siddharth et al. (2017) ; Kingma et al. (2014) base on VAEs and obtain disentangled image representations with semi-supervised learning.  ... 
arXiv:1703.00955v4 fatcat:rcitmftuzvh6vasqalw63jr23a

Evaluating the Interpretability of Generative Models by Interactive Reconstruction [article]

Andrew Slavin Ross, Nina Chen, Elisa Zhao Hang, Elena L. Glassman, Finale Doshi-Velez
2021 arXiv   pre-print
This is especially true in representation learning, where interpretability research has focused on "disentanglement" measures only applicable to synthetic datasets and not grounded in human factors.  ...  For machine learning models to be most useful in numerous sociotechnical systems, many have argued that they must be human-interpretable.  ...  ACKNOWLEDGMENTS We thank Krzysztof Gajos, Zana Buçinca, Zilin Ma, Giulia Zerbini, Benjamin Levy, and the Harvard DtAK group for helpful discussions and insights.  ... 
arXiv:2102.01264v1 fatcat:fdt2wt7abveplayhib5z5gevii

Weakly-Supervised Reinforcement Learning for Controllable Behavior [article]

Lisa Lee, Benjamin Eysenbach, Ruslan Salakhutdinov, Shixiang Shane Gu, Chelsea Finn
2020 arXiv   pre-print
On a variety of challenging, vision-based continuous control problems, our approach leads to substantial performance gains, particularly as the complexity of the environment grows.  ...  Reinforcement learning (RL) is a powerful framework for learning to take actions to solve tasks.  ...  We thank Michael Ahn for help with running the manipulation experiments, and also thank Benjamin Narin, Rafael Rafailov, and Riley DeHaan for help with initial experiments.  ... 
arXiv:2004.02860v2 fatcat:qjie7lfegzfbteknc6ywan5mb4

Variational Bandwidth Auto-encoder for Hybrid Recommender Systems [article]

Yaochen Zhu, Zhenzhong Chen
2021 arXiv   pre-print
In this article, we propose a variational bandwidth auto-encoder (VBAE) for recommendations, aiming to address the sparsity and noise problems simultaneously.  ...  control the information allowed to be accessed from the feature embeddings.  ...  ., ratings and user features in recommender systems.  ... 
arXiv:2105.07597v2 fatcat:eajcoprjefeorbr4qsklmwerwa

Weakly Supervised Disentangled Representation for Goal-conditioned Reinforcement Learning

Zhifeng Qian, You Mingyu, Zhou Hongjun, Bin He
2022 IEEE Robotics and Automation Letters  
Due to the high controllability of the representations, STAE can simply recombine and recode the representations to generate unseen goals for agents to practice themselves.  ...  In the paper, we propose a skill learning framework DR-GRL that aims to improve the sample efficiency and policy generalization by combining the Disentangled Representation learning and Goal-conditioned  ...  Representation Learning for Vision-based control For vision-based control tasks, many related works aim to learn an expressive representation to improve the sample efficiency of visual RL. Park et al.  ... 
doi:10.1109/lra.2022.3141148 fatcat:sprl5ju4x5ftbmq76cp3c3cai4

Trajectory-User Linking via Variational AutoEncoder

Fan Zhou, Qiang Gao, Goce Trajcevski, Kunpeng Zhang, Ting Zhong, Fengli Zhang
2018 Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence  
Trajectory-User Linking (TUL) is an essential task in Geo-tagged social media (GTSM) applications, enabling personalized Point of Interest (POI) recommendation and activity identification.  ...  We tackle the TUL problem with a semi-supervised learning framework, called TULVAE (TUL via Variational AutoEncoder), which learns the human mobility in a neural generative architecture with stochastic  ...  Science Foundation of China (Grant No.61602097, No.61472064 and No.61502087), NSF grants III 1213038 and CNS 1646107, ONR grant N00014-14-10215 and HERE grant 30046005, and the Fundamental Research Funds for  ... 
doi:10.24963/ijcai.2018/446 dblp:conf/ijcai/0002GTZZZ18 fatcat:nygjb5747rf3nect5clx46higu

Autoencoders [article]

Dor Bank, Noam Koenigstein, Raja Giryes
2021 arXiv   pre-print
input is similar as possible to the original one.  ...  An autoencoder is a specific type of a neural network, which is mainly designed to encode the input into a compressed and meaningful representation, and then decode it back such that the reconstructed  ...  A classical approach in recommender system models is Collaborative Filtering (CF) [22] . In CF, user preferences are inferred based on information from other user preferences.  ... 
arXiv:2003.05991v2 fatcat:qs45nliutfc4zmkfpzditgn464

A Disentangled Adversarial Neural Topic Model for Separating Opinions from Plots in User Reviews [article]

Gabriele Pergola, Lin Gui, Yulan He
2020 arXiv   pre-print
Although these approaches have achieved significant results, surprisingly very little work has been done on how to disentangle the latent topics.  ...  In this paper, we propose a neural topic model combined with adversarial training to disentangle opinion topics from plot and neutral ones.  ...  that can be controlled exploiting a reinforcement learning mechanism determining the disentangled factors.  ... 
arXiv:2010.11384v1 fatcat:ognbedruwva4bgcvbkqol22gwq

Multi-Auxiliary Augmented Collaborative Variational Auto-encoder for Tag Recommendation [article]

Jing Yi, Xubin Ren, Zhenzhong Chen
2022 arXiv   pre-print
Specifically, the model learns deep latent embeddings from different item auxiliary information using variational auto-encoders (VAE), which could form a generative distribution over each auxiliary information  ...  information and content information for better recommendations.  ...  MacridVAE [29] extended VAE-CF by learning disentangled representations of users' latent preferences at high (e.g., a user's intention towards different categories of items) and low (e.g., a user's preference  ... 
arXiv:2204.09422v1 fatcat:zozcuc526fbjxp4d5t7kah3t24
« Previous Showing results 1 — 15 out of 273 results