[논문]오토인코더를 이용한 딥러닝 기반 추천시스템 모형의 비교 연구

이효진; 정윤서

doi:10.5351/kjas.2021.34.3.329

오토인코더를 이용한 딥러닝 기반 추천시스템 모형의 비교 연구
Comparison of deep learning-based autoencoders for recommender systems 원문보기

응용통계연구 = The Korean journal of applied statistics, v.34 no.3, 2021년, pp.329 - 345

이효진 (고려대학교 통계학과) , 정윤서 (고려대학교 통계학과)

초록
AI-Helper

추천 시스템은 고객의 데이터를 이용하여 개인 맞춤화된 상품을 추천한다. 추천 시스템은 협업 필터링, 콘텐츠 기반 필터링 그리고 이 두 가지를 합친 하이브리드 방법의 세 가지로 크게 나누어진다. 이 연구에서는 딥러닝 방법론에 기초한 오토인코더를 이용한 추천 시스템에 대한 소개와 그 모형들의 비교 연구를 진행한다. 오토인코더는 데이터 행렬에 0이 많은 경우의 문제를 효과적으로 다룰 수 있는 딥러닝 기반의 비지도학습 모형이다. 이 연구에서는 세 개의 실제 데이터를 이용하여 다섯 가지 종류의 오토인코더 기반 모형들을 비교한다. 처음의 세 개 모형은 협업 필터링에 속한 모형이고 나머지 두 개의 모형은 하이브리드 모형이다. 실제 데이터는 고객의 평점 데이터이고, 대부분의 평점이 없어서 희박성 비율이 높다는 특징이 있다.

Abstract ▼ AI-Helper

Recommender systems use data from customers to suggest personalized products. The recommender systems can be categorized into three cases; collaborative filtering, contents-based filtering, and hybrid recommender system that combines the first two filtering methods. In this work, we introduce and compare deep learning-based recommender system using autoencoder. Autoencoder is an unsupervised deep learning that can effective solve the problem of sparsity in the data matrix. Five versions of autoencoder-based deep learning models are compared via three real data sets. The first three methods are collaborative filtering and the others are hybrid methods. The data sets are composed of customers' ratings having integer values from one to five. The three data sets are sparse data matrix with many zeroes due to non-responses.

주제어

표/그림 (10)

그림 Figure 1: The structure of autoencoder consisting of encoder and decoder considering a non-linear relationship as an activation function.
$Figure 2: The structure of denoising autoencoder containing input value with added noise <TEX>$\tilde{x}$</TEX>, which is distinct from the basic autoencoder.$ 그림 Figure 2: The structure of denoising autoencoder containing input value with added noise $\tilde{x}$, which is distinct from the basic autoencoder.
$Figure 3: Understanding denoising autoencoder from manifold learning perspective (Vincent et al. 2008). <TEX>$\tilde{x}$</TEX>i is generated from qD(<TEX>$\tilde{x}$</TEX>i|xi) and the denoising autoencoder tries to closely map g𝜃' ( f𝜃(<TEX>$\tilde{x}$</TEX>i)) to real input value xi.$ 그림 Figure 3: Understanding denoising autoencoder from manifold learning perspective (Vincent et al. 2008). $\tilde{x}$_i is generated from q_D($\tilde{x}$_i|x_i) and the denoising autoencoder tries to closely map g_𝜃' ( f_𝜃($\tilde{x}$_i)) to real input value x_i.
그림 Figure 4: The structure of variational autoencoder.
표 Table 1: Mean of masked RMSE of autoencoder and denoising autoencoder and Kullback-Leibler divergence of variational autoencoder on MovieLens 1M data set. For each model, two activation functions are used. The first one is the activation function for each layer, and the second one is for the last layer before an output layer
표 Table 2: Mean of masked RMSE of autoencoder and denoising autoencoder and Kullback-Leibler divergence of variational autoencoder on Amazon Review data set. For each model, two activation functions are used. The first one is the activation function for each layer, and the second one is for the last layer before an output layer
표 Table 3: Mean of masked RMSE of autoencoder and denoising autoencoder, and Kullback-Leibler divergence of variational autoencoder on Yelp Review data set. For each model, two activation functions are used. The first one is the activation function for each layer, and the second one is for the last layer before an output layer
표 Table 4: Mean of masked RMSE of contractive autoencoder on the three data sets at various hyperparameter settings. For each model, two activation functions are used. The first one is the activation function for each layer, and the second one is for the last layer before an output layer
표 Table 5: Mean of masked RMSE of AutoSVD and AutoSVD++ on the three data sets at various hyperparameter values
표 Table 6: Masked RMSE and MAE of the five models with 'optimally' tuned parameters on the three data sets. For each model, two activation functions are used. The first one is the activation function for each layer, and the second one is for the last layer before an output layer

참고문헌 (17)

Aggarwal CC (2018). Neural Networks and Deep Learning, Springer, NewYork.
Ali SM, Nayak GK, Lenka RK, and Barik RK (2018). Movie recommendation system using genome tags and content-based filtering, Advances in Data and Information Sciences(pp.85-94), Springer, NewYork.
Harper FM and Konstan JA (2015). The movielens datasets: History and context, ACM Transactions on Interactive Intelligent Systems, 5, 1-19.

상세보기
Kingma DP and Welling M (2013). Auto-Encoding Variational Bayes, arXiv preprint arXIV:1312.6114.
Koren Y (2008). Factorization meets the neighborhood: a multifaceted collaborative filtering model. In Proceedings of the 14th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 426-434.
Koren Y, Bell R, and Volinsky C (2009). Matrix factorization techniques for recommender systems, Computer, 42, 30-37.

상세보기
Kuchaiev O and Ginsburg B (2017). Training Deep Autoencoders for Collaborative Filtering, arXiv preprint arXiv:1708.01715.
Ni J, Li J, and McAuley J (2019). Justifying recommendations using distantly-labeled reviews and fine-grained aspects. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), 188-197.
Odaibo S (2019). Tutorial: Deriving the Standard Variational Autoencoder (vae) Loss Function, arXiv preprint arXiv:1907.08956.
Rifai S, Vincent P, Muller X, Glorot X, and Bengio Y (2011). Contractive auto-encoders: Explicit invariance during feature extraction. In Proceedings of the 28th International Conference on Machine Learning June 2011, 833-840
Schafer JB, Frankowski D, Herlocker J, and Sen S (2007). Collaborative filtering recommender systems, The Adaptive Web(pp. 291-324), Springer, NewYork.
Sedhain S, Menon AK, Sanner S, and Xie L (2015). Autorec: Autoencoders meet collaborative filtering. In Proceedings of the 24th international conference on World Wide Web, 111-112.
Vincent P, Larochelle H, Bengio Y, and Manzagol PA (2008). Extracting and composing robust features with denoising autoencoders. In Proceedings of the 25th International Conference on Machine Learning, 1096-1103.
Wu Y, DuBois C, Zheng AX, and Ester M (2016). Collaborative denoising auto-encoders for top-n recommender systems. In Proceedings of the Ninth ACM International Conference on Web Search and Data Mining, 153-162.
Zhang S, Yao L, Sun A, and Tay Y (2019). Deep learning based recommender system: A survey and new perspectives, ACM Computing Surveys (CSUR), 52, 1-38.
Zhang S, Yao L, and Xu X (2017). Autosvd++ an efficient hybrid collaborative filtering model via contractive auto-encoders. In Proceedings of the 40th International ACM SIGIR Conference on Research and Development in Information Retrieval, 957-960.
Zheng N and Xue J (2009). Manifold Learning(pp.87-119), Springer, London.

표제어: PCR

동의어: Packet Collision Rate

용어 설명 출처 목록 (6)

용어 설명: PCR은 세균 특이성이 있는 primer를 이용하여 적은 수의 세균이 있을지라도 쉽게 검출할 수 있는 유용한 방법이며, 이를 이용하여 구강 내 치면세균막이나 타액에서 직접 세균을 검출할 수 있게 되었다[8].

내보내기 구분	파일저장 인쇄 메일전송
구성항목	기본정보 상세정보 관리번호, 논문명, 저널/프로시딩명, 저자 , 발행년, 권, 호, 시작페이지, 끝페이지, 발행기관 관리번호, 논문명, 대등논문명, 저자 , 저널/프로시딩명, 발행기관, 발행년, 발행언어, 권, 호, 시작페이지, 끝페이지, ISBN, ISSN, 주제분야, 키워드, 초록(한글), 초록(영문), 저자(소속기관)
저장형식	Text(ASCII format) Excel format RefWorks Direct Export RIS format (for Reference Manager, ProCite, EndNote), Scholar's Aids, Mendeley
메일정보	받는사람 (필수) @ 보내는사람 (선택) @ 제목 내용 KISTI 검색결과 이메일 서비스
안내	총 건의 자료가 검색되었습니다. 다운받으실 자료의 인덱스를 입력하세요. (1-10,000) 검색결과의 순서대로 최대 10,000건 까지 다운로드가 가능합니다. 데이타가 많을 경우 속도가 느려질 수 있습니다.(최대 2~3분 소요) 다운로드 파일은 UTF-8 형태로 저장됩니다. 파일의 내용이 제대로 보이지 않을실 때는 웹브라우저 상단의 보기 -> 인코딩 -> 자동선택 여부를 확인하십시오. ~ Text(ASCII format) Excel format

연합인증