精选两篇最新AutoML推荐系统综述

会员服务 ·

精选两篇最新AutoML推荐系统综述

2022 年 4 月 7 日 机器学习与推荐算法

收录两篇基于AutoML的推荐系统最新综述。

1. 基于AutoML的深度推荐系统：综述

Title: Automated Machine Learning for Deep Recommender Systems: A Survey

Published: 2022-04-04

Url: http://arxiv.org/abs/2204.01390v1

Authors: Bo Chen,Xiangyu Zhao,Yejing Wang,Wenqi Fan,Huifeng Guo,Ruiming Tang

深度推荐系统（Deep recommender systems，DRS）对于当前的商业在线服务提供商来说至关重要，后者通过推荐适合用户兴趣和偏好的项目来解决信息过载问题。它们具有前所未有的特征表示效率和建模用户和项目之间非线性关系的能力。尽管有了进步，DRS模型和其他深度学习模型一样，采用了复杂的神经网络架构和其他关键组件，这些组件通常由人类专家设计和调整。本文将对开发DRS模型的自动机器学习（AutoML）进行全面总结。Wefirst概述了DRS模型的AutoML和相关技术。然后，我们讨论了最先进的AutoML方法，这些方法可以自动化DRS中的特征选择、特征嵌入、特征交互和系统设计。最后，我们讨论了一些不错的研究方向，并总结了综述。

Deep recommender systems (DRS) are critical for current commercial onlineservice providers, which address the issue of information overload by recommending items that are tailored to the user's interests and preferences.They have unprecedented feature representations effectiveness and the capacityof modeling the non-linear relationships between users and items. Despite theiradvancements, DRS models, like other deep learning models, employ sophisticatedneural network architectures and other vital components that are typicallydesigned and tuned by human experts. This article will give a comprehensivesummary of automated machine learning (AutoML) for developing DRS models. Wefirst provide an overview of AutoML for DRS models and the related techniques.Then we discuss the state-of-the-art AutoML approaches that automate thefeature selection, feature embeddings, feature interactions, and system designin DRS. Finally, we discuss appealing research directions and summarize the survey.

来一览所有的AutoML Recsys：

2. 深度推荐系统的AutoML：综述

Title: AutoML for Deep Recommender Systems: A Survey

Published: 2022-03-25

Url: http://arxiv.org/abs/2203.13922v1

Authors: Ruiqi Zheng,Liang Qu,Bin Cui,Yuhui Shi,Hongzhi Yin

推荐系统在信息过滤中发挥着重要作用，并已被用于不同的场景，如电子商务和社交媒体。随着深度学习的蓬勃发展，深度推荐系统通过捕捉非线性信息和项目用户关系显示出优越的性能。然而，深度推荐系统的设计在很大程度上依赖于人类经验和专家知识。为了解决这个问题，引入了自动机器学习（AutoML）来自动搜索深度推荐系统不同部分的适当条件。本调查对该领域的文献进行了全面回顾。首先，我们为深度推荐系统（AutoRecSys）提出了一个抽象的AutoML概念，描述了它的构建模块，并将其与传统的AutoML技术和推荐系统区分开来。其次，我们将共济失调作为一个分类框架，包括嵌入维度搜索、特征交互搜索、模型设计搜索和其他组件搜索。此外，我们特别强调搜索空间和搜索策略，因为它们是连接每个类别中所有方法的共同线索，使从业者能够分析和比较各种方法。最后，我们提出了四个未来有希望的研究方向，将引领这一研究方向。

Recommender systems play a significant role in information filtering and havebeen utilised in different scenarios, such as e-commerce and social media. Withthe prosperity of deep learning, deep recommender systems show superiorperformance by capturing non-linear information and item-user relationships.However, the design of deep recommender systems heavily relies on humanexperiences and expert knowledge. To tackle this problem, Automated MachineLearning (AutoML) is introduced to automatically search for the propercandidates for different parts of deep recommender systems. This surveyperforms a comprehensive review of the literature in this field. Firstly, wepropose an abstract concept for AutoML for deep recommender systems(AutoRecSys) that describes its building blocks and distinguishes it fromconventional AutoML techniques and recommender systems. Secondly, we present ataxonomy as a classification framework containing embedding dimension search,feature interaction search, model design search and other components search.Furthermore, we put a particular emphasis on the search space and searchstrategy, as they are the common thread to connect all methods within eachcategory and enable practitioners to analyse and compare various approaches.Finally, we propose four future promising research directions that will leadthis line of research.