O’REILLY、INTEL AI主办

English中文
将人工智能用起来
2019年6月18-21日
北京,中国

Real-time product recommendations leveraging deep learning on Apache Spark in Office Depot

此演讲使用中文 (This will be presented in Chinese)

Guoqiong Song (Intel), Luyang Wang (Office Depot), Jiao(Jennie) Wang (Intel), Jing (Nicole) Kong (Office Depot)
14:5015:30 Thursday, June 20, 2019
实施人工智能 (Implementing AI), 模型与方法 (Models and Methods)
Location: 多功能厅5A+B(Function Room 5A+B)

必要预备知识 (Prerequisite Knowledge)

  • A basic understanding of Apache Spark, machine learning, and deep learning

您将学到什么 (What you'll learn)

  • Learn about a run-time recommendation system built with deep learning using Analytics Zoo on BigDL and Apache Spark
  • Gain insight into the process for developing a full end-to-end deep learning workflow including elements of big data and machine learning in the cloud

描述 (Description)

Real-time recommender systems are critical for the success of the ecommerce industry. Newly developed deep neural networks (DNNs) have shown success as recommender systems by capturing the nonlinear relationships in the user-item dataset.

Guoqiong Song, Luyang Wang, Jiao Wang, and Jing Kong showcase how to build efficient real-time recommender systems to recommend products for different users by leveraging different types of DNNs. They use Analytics Zoo for Spark and BigDL to build the end-to-end flow on AWS at scale.

The system first processes all the transaction data and extract features on AWS; then it trains a comprehensive recommender model, including neural collaborative filtering network, wide and deep network, and a session-based recommender with recurrent neural networks. The system further serves the recommender model for each user on web service. By adopting the end-to-end flow of the Analytics Zoo solution, you can see a big improvement of accuracy compared to traditional recommendation algorithms.

实时推荐系统对电子商务行业的成功至关重要。近年来出现的深度神经网络(DNN)通过捕获用户—商品数据集中的非线性关系,已经在推荐系统中获得了成功。 本议题将说明如何构建高效的实时推荐系统,来利用不同类型的DNN为不同用户推荐产品。

本次议题将解释如何使用基于Spark和BigDL的Analytics Zoo在AWS上大规模构建端到端流程。整个系统首先处理AWS上的所有交易数据并提取特征。然后,它训练了一个完整的推荐模型,包括神经协同过滤网络、宽且深的网络,以及基于会话的循环神经网络推荐器。这个系统进一步提供了推荐模型的服务,为网站上的每个用户提供推荐。通过采用Analytics Zoo解决方案的端到端流程,我们看到与传统推荐算法相比准确性有了大幅提升。

Photo of Guoqiong Song

Guoqiong Song

Intel

Guoqiong Song is a senior deep learning software engineer of the big data technology team at Intel. Her interest is in developing and optimizing distributed deep learning algorithms on Spark. She has a PhD in atmospheric and oceanic sciences from UCLA with a focus on numerical modeling and optimization.

Guoqiong Song是英特尔大数据技术团队的高级深度学习软件工程师。 她拥有加州大学洛杉矶分校的大气和海洋科学博士学位,专业方向是数值建模和优化。 她现在的研究兴趣是开发和优化分布式深度学习算法。

Photo of Luyang Wang

Luyang Wang

Office Depot

Lu Wang is a data scientist and big data engineer from Office Depot, where he works on machine learning and big data analytics. He’s engaged in developing distributed machine learning applications and real-time web services for Office Depot digital business platform.

Office Depot数据科学家/大数据工程师,负责机器学习和大数据分析。他致力于为Office Depot数字业务平台开发分布式机器学习应用和实时Web服务。

Photo of Jiao(Jennie) Wang

Jiao(Jennie) Wang

Intel

Jiao (Jennie) Wang is a software engineer on the big data technology team at Intel, where she works in the area of big data analytics. She’s engaged in developing and optimizing distributed deep learning framework on Apache Spark.

Jiao(Jennie)Wang是英特尔大数据技术团队的软件工程师,主要工作在大数据分析领域。她致力于基于Apache Spark开发和优化分布式深度学习框架。

Photo of Jing (Nicole) Kong

Jing (Nicole) Kong

Office Depot

Jing (Nicole) Kong is a data scientist at Office Depot, where she deals with big data and transforms data and models into products and service that drive business. She’s experienced with a number of different machine learning and deep learning models.

Jing(Nicole)是一位数据科学家,拥有构建不同的机器学习/深度学习模型、处理大数据以及将数据/模型转换为推动业务发展的产品和服务的经验。

Leave a Comment or Question

Help us make this conference the best it can be for you. Have questions you'd like this speaker to address? Suggestions for issues that deserve extra attention? Feedback that you'd like to share with the speaker and other attendees?

Join the conversation here (requires login)