Recurrent Recommender Networks

Wu, Chao-Yuan; Ahmed, Amr; Beutel, Alex; Smola, Alexander J.; Jing, How

doi:10.1145/3018661.3018689

Cited by 544 publications

(254 citation statements)

References 14 publications

Supporting

Mentioning

250

Contrasting

Unclassified

Order By: Relevance

“…[48] took one-hot encoding of items in users' behavior sequences as input, converted them into dense vectors via lookup operation, which are sequentially fed into GRUbased RNN to learn users' historical embedding. Finally, the model projects user u's historical embedding h, user embedding, and target item i's embedding to predict user u's probability in choosing item i. RRN [49] is the first recurrent recommender network that attempts to capture the dynamics of both user and item representation, where each individual recurrent network is adopted to address the temporal evolution of each user and item respectively. To capture stationary attributes, it uses an additional set of auxiliary parameters for users and items respectively.…”

Section: Rnn-based Modelsmentioning

confidence: 99%

“…User recurrent models. They treat both user and item representations as recurrent components in DL-based models, which can better capture users' evolving preferences, including memory-augmented neural network [51], RNNbased models [52], [53], [74] and recurrent neural networks [49], [50]. For example, [52], [53], [74] use RNN framework to learn users' long-term interest from their historical behavior sequences, and experiments verify that considering a user's long-term interest is critically valuable for personalized recommendation, e.g., HRNN [52] exceeds GRU4Rec by 3.5% with user representation.…”

Section: Adding Explicit User Representationmentioning

confidence: 99%

See 1 more Smart Citation

Deep Learning-Based Sequential Recommender Systems: Concepts, Algorithms, and Evaluations

Fang

Guo

Zhang

et al. 2019

Lecture Notes in Computer Science

View full text Add to dashboard Cite

In the field of sequential recommendation, deep learning methods have received a lot of attention in the past few years and surpassed traditional models such as Markov chain-based and factorization-based ones. However, DL-based methods also have some critical drawbacks, such as insufficient modeling of user representation and ignoring to distinguish the different types of interactions (i.e., user behavior) among users and items. In this view, this survey focuses on DL-based sequential recommender systems by taking the aforementioned issues into consideration. Specifically, we illustrate the concept of sequential recommendation, propose a categorization of existing algorithms in terms of three types of behavioral sequence, summarize the key factors affecting the performance of DL-based models, and conduct corresponding evaluations to demonstrate the effects of these factors. We conclude this survey by systematically outlining future directions and challenges in this field.Index Terms-sequential recommendation, session-based recommendation, sequential data, deep learning, influential factors ! • Hui Fang is with the 1. We searched on arXiv.org with keywords related to the sequential recommendation and DL techniques in March 2019.2. Fig. 3: A schematic diagram of the sequence recommendation. c i : behavior type, o i : behavior object. A behavior is represented by a 2-tuple. A behavior sequence (i.e., behavior trajectory) is a time ordered list of 2-tuples.The input behavior sequence {a 1 , a 2 , a 3 , ..., a t } is polymorphic, which can thus be divided into three types: experience-based, transaction-based and interaction-based behavior sequence, whose details are introduced as follows:

show abstract

Section: Rnn-based Modelsmentioning

confidence: 99%

Section: Adding Explicit User Representationmentioning

confidence: 99%

Deep Learning-Based Sequential Recommender Systems: Concepts, Algorithms, and Evaluations

Fang

Guo

Zhang

et al. 2019

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…files and reviews [3], [5]. In particular, deep learning models have been widely studied [13], [15]. AutoRec first proposed the use of autoencoders for recommender systems [17].…”

Section: Figmentioning

confidence: 99%

Deep Heterogeneous Autoencoders for Collaborative Filtering

Stenger

et al. 2018

2018 IEEE International Conference on Data Mining (ICDM)

View full text Add to dashboard Cite

This paper leverages heterogeneous auxiliary information to address the data sparsity problem of recommender systems. We propose a model that learns a shared feature space from heterogeneous data, such as item descriptions, product tags and online purchase history, to obtain better predictions. Our model consists of autoencoders, not only for numerical and categorical data, but also for sequential data, which enables capturing user tastes, item characteristics and the recent dynamics of user preference. We learn the autoencoder architecture for each data source independently in order to better model their statistical properties. Our evaluation on two MovieLens datasets and an ecommerce dataset shows that mean average precision and recall improve over state-of-the-art methods.

show abstract

“…Uma Rede Neural Recorrente (RNN)é uma classe de redes neurais que exploram a natureza sequencial da sua entrada [Wu et al 2017]. As RNNs têm sido aplicadas em várias tarefas como: reconhecimento de dígitos [Graves and Schmidhuber 2009], geração de texto [Sutskever et al 2011], criação de modelos de língua [Mikolov 2012, Graves 2013, Pascanu et al 2013, reconhecimento de voz [Graves et al 2013], tradução automática [Bahdanau et al 2014, Sutskever et al 2014] e inferência textual [Bowman et al 2015, Rocktäschel et al 2015.…”

Section: Rede Neural Recorrente E a Variante Long Short Term Memoryunclassified

Um Método para Avaliação Automática da Qualidade de Recursos Educacionais Abertos usando Deep Learning

Gazzola¹

2017

Anais Do XXVIII Simpósio Brasileiro De Informática Na Educação (SBIE 2017)

View full text Add to dashboard Cite

Abstract. Open Educational Resources (OER) are documents that are openly licensed and used for teaching, learning, and research purposes. They cover complete courses, textbooks, videos, softwares and any other tools, materials or techniques to support access to knowledge. The main difficulty, however, is to ensure the quality of these educational resources stored in online repositories. To fill this gap, a method was created using deep neural networks, specifically, a Recurrent Neural Network (RNN) to evaluation the quality of open educational resources, and compared with a Supporting Vector Machine (SVM) and its variations. The research methodology used was the use of an architecture for neural network, the creation of a controlled scenario, and the comparison with the main studies that perform automated evaluation of OER.Resumo. Recursos Educacionais Abertos (REAs) são documentos abertamente licenciados e usados para fins de ensino, aprendizagem e pesquisa. Abrangem cursos completos, livros didáticos, vídeos, softwares e quaisquer outras ferramentas, materiais ou técnicas para apoiar o acesso ao conhecimento. A principal dificuldade, porém,é garantir a qualidade desses recursos educacionais armazenados em repositórios on-line. Para preencher esta lacuna, foi criado um método usando redes neurais profundas, especificamente, uma Rede Neural Recorrente (RNN) para avaliação automatizada da qualidade de recursos educacionais abertos, sendo comparado com uma Máquina de Vetores de Suporte (SVM) e suas variações. A metodologia de pesquisa utilizada foi a criação de uma arquitetura para rede neural, a criação de um cenário controlado, e a comparação com os principais trabalhos que realizam avaliação automatizada de REAs. IntroduçãoRecursos Educacionais Abertos (REAs) foram definidos por diversos trabalhos na literatura, sendo objeto de estudo de uma ampla diversidade de trabalhos [Wiley et al. 2014]. O termo foi cunhado em 2002 pela Unesco [UNESCO. 2002] que convocou o fórum sobre o impacto dos cursos abertos do ensino superior em países em desenvolvimento. Foi definido neste fórum que o termo "aberto" relacionava com recursos educacionais e indicava as possibilidades do conhecimento ser consultado, usado e adaptado por todos sem fins comerciais.A ideia dos REAsé tornar o conhecimento do mundo um bem público tendo a tecnologia em geral, e especificamente a Web, como uma fonte de acesso a esse conhecimento. Entende-se, assim, que os REAs devem permitir seu compartilhamento, uso e

show abstract

Recurrent Recommender Networks

Cited by 544 publications

References 14 publications

Deep Learning-Based Sequential Recommender Systems: Concepts, Algorithms, and Evaluations

Deep Learning-Based Sequential Recommender Systems: Concepts, Algorithms, and Evaluations

Deep Heterogeneous Autoencoders for Collaborative Filtering

Um Método para Avaliação Automática da Qualidade de Recursos Educacionais Abertos usando Deep Learning

Contact Info

Product

Resources

About