Applying Deep Learning To Airbnb Search Unique Challenges High - PowerPoint PPT Presentation

Jan 03, 2024 •297 likes •622 views

MALAY HALDAR / April 2019 / QCon Applying Deep Learning To Airbnb Search Unique Challenges High Value Transactions Sparse Listing Data Sparse User Data Users are picky when spending Each listing can be booked at Travel is not a daily use

MALAY HALDAR / April 2019 / QCon Applying Deep Learning To Airbnb Search
Unique Challenges High Value Transactions Sparse Listing Data Sparse User Data Users are picky when spending Each listing can be booked at Travel is not a daily use case. hundreds of dollars. most 365 times a year.
How Does Airbnb Search Work?
Evolution Of Search Ranking At Airbnb DNN GBDT Scoring Functions 2014 2015 2016 2017 2018 2019
First Neural Net : April 2017 L2 loss 31 Relus { Continuous features with heavy feature engineering }
● Keep the first launch as simple as possible Key Learnings ● Manage expectations, aim for neutral
Second Neural Net: Lambdarank NN June 2017 sigmoid_cross_entropy_with_logits X lambdarank_weights - { Booked listing features } { Unbooked listing features }
● Pairwise preference Key Learnings ● Tensorflow
Third Neural Net: Hybrid Mar 2018
DNN : June 2018 83 Relus 127 Relus { Raw query and listing stats }
● Transitioning to deep learning is about scaling the system. Key Learnings ● Complexity is easy to build, simplicity is hard.
Gains
Problem Formulation
Pointwise
Pairwise
Do Not Cook Your Features Feature Engineering
Continuous Features
Embedding -1.203 -1.862 Categorical 1.701 Feature -2.003 ( 0x3fc90 ) 0.192 : Categorical Features
● I/O System ● I/O Engineering ● I/O
https://christophm.github.io/interpretable-ml-book/pdp.html ...This problem is easily solved by showing a rug (indicators for data points on the x-axis) or a histogram. Feature The assumption of independence is the biggest Importance issue with PD plots. It is assumed that the feature(s) for which the partial dependence is computed are not correlated with other features. For example, suppose you want to predict ...
TopBot Analysis
Hyperparameter Optional, not a necessity. Tuning
Many more interesting Failed models! topics...
Applying Deep Learning To Airbnb Search Malay Haldar, Mustafa Abdool, Prashant Ramanathan, Tao Xu, Shulin Yang, Huizhong Duan, Qing Zhang, Nick Barrow-Williams, Bradley C. Turnbull, Brendan M. Collins, Thomas Legrand https://arxiv.org/abs/1810.09591

Download Presentation

Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend

Bighead Airbnbs End-to-End Machine Learning Infrastructure Andrew Hoh ML Infra @ Airbnb

Bighead Airbnbs End-to-End Machine Learning Infrastructure Andrew Hoh ML Infra @ Airbnb Architecture Background Design Goals Open Source Deep Dive Background Airbnbs Product A global travel community that offers magical

1.06k views • 66 slides

Deep Reinforcement Learning [Mastering the Game of Go with Deep Reinforcement Learning and Tree

Deep Reinforcement Learning [Mastering the Game of Go with Deep Reinforcement Learning and Tree Search, Nature 2016] CS 486/686 University of Waterloo Lecture 21: July 12, 2017 Outline AlphaGo Supervised Learning of Policy Networks

540 views • 15 slides

Public Policy and Deep Reinforcement Learning on AWS Emily Webber | Machine Learning Specialist

Public Policy and Deep Reinforcement Learning on AWS Emily Webber | Machine Learning Specialist at Amazon Web Services | To-be-open-sourced research project Public Policy Has Unique Challenges Structural Inefficiency Lack of single goal

387 views • 20 slides

Applying Deep Learning to Financial Markets with News Data Executive Summary Large amounts of

Applying Deep Learning to Financial Markets with News Data Executive Summary Large amounts of Use deep neural Predict market news data networks to build direction of predictive models relevant stocks Dealing with News Original Workflow

480 views • 7 slides

Applying Deep Neural Network Techniques for LArTPC Data Reconstruction Laura Domine (Stanford

Applying Deep Neural Network Techniques for LArTPC Data Reconstruction Laura Domine (Stanford University / SLAC) Fermilab Machine Learning Group meeting - 11/7/18 Plan 1. LArTPC & Deep Learning 2. Examples of applications: UResNet &

585 views • 39 slides

Realize your definition of success. A deep dive on three unique marketers and their paths to

Realize your definition of success. A deep dive on three unique marketers and their paths to achieving their goals. No two marketers are alike. Every brand faces unique challenges. And all brands set their own unique marketing goals to

156 views • 14 slides

Deep Learning for Semantic Search in E-commerce Somnath Banerjee Head of Search Algorithms at

Deep Learning for Semantic Search in E-commerce Somnath Banerjee Head of Search Algorithms at Walmart Labs https://www.linkedin.com/in/somnath-banerjee/ March 19, 2019 Walmart E-commerce search problem E-commerce Search Store Associate

510 views • 46 slides

Op#miza#on Challenges for Deep Learning Yoshua Bengio U.

Op#miza#on Challenges for Deep Learning Yoshua Bengio U. Montreal December 12th, 2014 OPT2014: NIPS Workshop on OpBmizaBon for Machine Learning

565 views • 45 slides

Airbnb Sustainable tourism for rural development Airbnb in Europe From urban to rural tourism

Sarah Prot - February 2017 Airbnb Sustainable tourism for rural development Airbnb in Europe From urban to rural tourism Airbnb helps to promote tourism in all regions of Europe Provence, France Andalucia, Spain Umbria, Italy Loch Lomond,

450 views • 16 slides

CS 285 Instructor: Sergey Levine UC Berkeley Challenges in Deep Reinforcement Learning Whats

Challenges and Open Problems CS 285 Instructor: Sergey Levine UC Berkeley Challenges in Deep Reinforcement Learning Whats the problem? Challenges with core algorithms : Stability: does your algorithm converge? Efficiency: how long

831 views • 31 slides

Structured Probabilistic Models for Deep Learning Lecture slides for Chapter 16 of Deep Learning

Structured Probabilistic Models for Deep Learning Lecture slides for Chapter 16 of Deep Learning www.deeplearningbook.org Ian Goodfellow 2016-10-04 Roadmap Challenges of Unstructured Modeling Using Graphs to Describe Model Structure

974 views • 34 slides

Regional Impacts of Airbnb L A U R A H O D G S O N P H D S C H O L A R A R C L I N K A G E

Regional Impacts of Airbnb L A U R A H O D G S O N P H D S C H O L A R A R C L I N K A G E P R O J E C T: L P 1 5 0 1 0 0 1 6 0 Airbnb a household name Short Term Rental Accommodation (STRA) Disintermediation Disruptor

227 views • 10 slides

Challenges in Applying Machine Learning Methods: Studying Political Interactions on Social

Challenges in Applying Machine Learning Methods: Studying Political Interactions on Social Networks CHAYA LIEBESKIND* AND KARINE NAHON ~ *Jerusalem College of Technology, Lev Academic Center, Jerusalem/Israel ~ Interdisciplinary Center Herzliya,

630 views • 15 slides

Challenges in Vessel Behavior and Anomaly Detection: From Classical Machine Learning to Deep

Challenges in Vessel Behavior and Anomaly Detection: From Classical Machine Learning to Deep Learning Lucas May Petry 1 , Amilcar Soares 2 , Vania Bogorny 1 , Bruno Brandoli 2 , Stan Matwin 2 1 Programa de Ps-Graduao em Cincias da

1.26k views • 16 slides

Deep learning: Challenges in learning and generalization Tomas Mikolov, Facebook AI What is

Deep learning: Challenges in learning and generalization Tomas Mikolov, Facebook AI What is generalization? Memorization to remember all training examples (Universal Approximator) 2 + 3 = 5, 3 + 2 = 5, ... Generalization to

281 views • 14 slides

Deep Reinforcement Learning 1 Outline 1. Overview of Reinforcement Learning 2. Policy Search 3.

Deep Reinforcement Learning 1 Outline 1. Overview of Reinforcement Learning 2. Policy Search 3. Policy Gradient and Gradient Estimators 4. Q-prop: Sample Efficient Policy Gradient and an Off-policy Critic 5. Model Based Planning in Discrete

769 views • 53 slides

Networking @ OLN 21 January 2017 ENGLISH LANGUAGE DESIRED OUTCOMES FOR EL PROFICIENCY IN

CHIJ Our Lady of the Nativity Networking @ OLN 21 January 2017 ENGLISH LANGUAGE DESIRED OUTCOMES FOR EL PROFICIENCY IN SINGAPORE All students to be able to use English to express themselves. All students to attain foundational skills,

510 views • 48 slides

Augurative and Commemorative Names from Early 14th-Century Imola Sara L. Uckelman @SaraLUckelman

Augurative and Commemorative Names from Early 14th-Century Imola Sara L. Uckelman @SaraLUckelman @thedmnes Institute for Medieval and Early Modern Studies Durham University 25th International Medieval Congress 04 July 2018, Leeds, England

258 views • 21 slides

Welcome to 3rd Grade Mrs. Szczureks Class Room 310 a r Y e o l h o S c 7 0 1 -

Welcome to 3rd Grade Mrs. Szczureks Class Room 310 a r Y e o l h o S c 7 0 1 - 2 1 6 2 0 About Mrs.Szczurek Years teaching: 12 years, ranging from early childhood to middle school grades Years teaching at Forest

676 views • 22 slides

What is a Book Parade? 1. It is a creative way to share one of your favorite books. 2. Explains

What is a Book Parade? 1. It is a creative way to share one of your favorite books. 2. Explains the parts of your book. 3. When done well, it makes someone else want to read the book. 4. Its kind of like a science fair, where you create a

585 views • 20 slides

Applications of GANs Photo-Realistic Single Image Super-Resolution Using a Generative

Applications of GANs Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks Generative Adversarial Text to Image Synthesis

666 views • 39 slides

I Can See Clearer Now The Blur is Gone Per Christian Hansen deblurring GAMM 2008 1 The

I Can See Clearer Now The Blur is Gone Per Christian Hansen deblurring GAMM 2008 1 The Speaker Per Christian Hansen Professor of Scientific Computing at DTU MSc EE 1982 PhD Num. Anal. 1985 Dr Techn 1996 Key

316 views • 27 slides

Modeling Semantic Overlap Over the last few years, broadening consensus that this is the core

Reasons to avoid Reasoning: Where does NLP stop and AI Begin? Bill Dolan Microsoft Research NSF symposium on Semantic Knowledge Discovery, Organization and Use November 15, 2008 Modeling Semantic Overlap Over the last few years, broadening

457 views • 20 slides

A Project Presented to The Faculty of Department of Computer Science San Jos State University

VIDEO SYNTHESIS from the StyleGAN Latent Space Video Synthesis from the StyleGAN Latent Space A Project Presented to The Faculty of Department of Computer Science San Jos State University Department of Computer Science San Jos State

770 views • 42 slides