SlideShare ist ein Scribd-Unternehmen logo
1 von 60
Downloaden Sie, um offline zu lesen
Recent Trends in
Personalization
at Netflix
Anuj Shah
@badshah79
https://www.linkedin.com/in/foranuj/
Why do we personalize?
Help members find entertainment
to watch and enjoy to maximize
member satisfaction and retention
Spark joy
What do we personalize?
Ordering of videos is personalized
From how we rank
Ranking
Selection and placement of rows is personalized
... to how we construct a page
Rows
... to how we respond to queries
Search query & result recommendation
... to how we cover different needs
Personalized instant choices
... to how we reach out
Message personalization
Everything is a recommendation!
Isn’t this solved yet?
○ Every person is unique with a variety of interests
… and sometimes multiple people use the same profile
○ Help people find what they want when they’re not sure what they want
○ Non-stationary, context-dependent, mood-dependent, ...
○ Large datasets but small data per member
… and potentially biased by the output of your system
○ Cold-start problems on all sides
○ More than just accuracy: diversity, novelty, freshness, fairness, ...
○ ...
No, personalization is hard!
So how are we going to solve this?
Some recent avenues in approaching these challenges:
1. Deep Learning
2. Causality
3. Bandits & Reinforcement Learning
4. Objectives
Trending Now
Trend 1: Deep Learning for
Recommendations
~2012 ~2017
Deep Learning
becomes popular in
Machine Learning
Deep Learning
becomes popular in
Recommender Systems
What took so long?
~2019
Traditional methods do
as well or better than
Deep Learning for
Recommender Systems
… Wait, what?
Timeline
Traditional Recommendations
Collaborative Filtering:
Recommend items that
similar users have chosen
0 1 0 1 0
0 0 1 1 0
1 0 0 1 1
0 1 0 0 0
0 0 0 0 1
Users
Items
U
≈
R
V
A Matrix Factorization view
2
U
A Feed-Forward Network view
V
2
U
A (deeper) feed-forward view
V
Mean
squared loss
?
… isn’t always the best
U
V
Mean squared
loss
?
Also see [Dacrema et al., 2019], [Rendle et. al,
2019], [Rendle et. al, 2021]. Make sure you
tune your baselines.
Understanding the relationships
From our forthcoming AI Magazine article “Deep Learning for Recommender Systems: A Netflix Case-Study”
X
R
EASE:
Embarrassingly Shallow Auto-Encoders [Steck, 2019]
● Super efficient model to train in a
collaborative filtering setting inspired
by SLIM
● Learn item-by-item matrix X such that
R.X is close to R and diag(X) is 0
○ Avoids trivial solution of identity
● Closed-form solution
● More on that: auto-encoders that
don’t overfit towards identity
≈ R
0
0
Modern Recommender Systems
0 1 0 1 0
0 0 1 1 0
1 0 0 1 1
0 1 0 0 0
0 0 0 0 1
Users
Items
Contextual event data
Modern Recommender Data
Model
Interactions
Impressions
Item data
Profile
settings
+
Contextual sequence data
2017-12-10 15:40:22
2017-12-23 19:32:10
2017-12-24 12:05:53
2017-12-27 22:40:22
2017-12-29 19:39:36
2017-12-30 20:42:13
Context Item
Sequence
per member
?
Time
V
Sequential Recommendation Network
Softmax
over items
Avg / Stack /
Sequence /
Attention
DNN / RNN / CNN / TNN
Input
interactions
(X)
(X)
p(Y|X)
2018-12-23
19:32:10
2018-12-24
12:05:53
2019-01-02
15:40:22
Offline Ranking Improvements
Trend 2: Causality
From Correlation to Causation
● Most recommendation algorithms
are correlational
○ Some early recommendation
algorithms literally computed
correlations between users and items
● Did you watch a movie because
we recommended it to you? Or
because you liked it? Or both?
● If you had to watch a movie, would
you like it? [Wang et al., 2020] p(Y|X) → p(Y|X, do(R))
(from http://www.tylervigen.com/spurious-correlations)
Feedback loops
Impression bias
inflates plays
Leads to inflated
item popularity
More plays
More
impressions
Oscillations in
distribution of genre
recommendations
Feedback loops can cause biases to be
reinforced by the recommendation system!
[Chaney et al., 2018]: simulations showing that this can reduce the
usefulness of the system
They’re real:
Lots of feedback loops...
Closed Loop
Training
Data
Watches Model
Recs
Search
Training
Data
Watches Model
Recs
Open Loop
Closed Loop
Training
Data
Watches Model
Recs
Danger Zone
Search
Training
Data
Watches Model
Recs
Open Loop
Closed Loop
Training
Data
Watches Model
Recs
Danger Zone
Search
Training
Data
Watches Model
Recs
Open Loop
V
Propensity Correction
Avg / Stack /
Sequence /
Attention
DNN / RNN / CNN / TNN
Input
interactions
(X)
(X)
2018-12-23
19:32:10
2018-12-24
12:05:53
2019-01-02
15:40:22
Policy
softmax
Propensity
softmax
E.g. [Chen et al., 2019]
p(Y|X, do(R))
p(R|X)
Challenges in Causal Recommendations
● Handling unobserved confounders
● Coming up with the right causal graph
● High variance (especially propensity-based ones)
● Computational challenges (e.g. [Wong, 2020])
● Off-policy evaluation
● When and how to introduce exploration
Trend 3: Bandits &
Reinforcement Learning in
Recommendations
Why contextual bandits for recommendations?
● Break feedback loops
● Want to explore to learn
● Uncertainty around member interests and new items
● Sparse and indirect feedback
● Changing trends
▶
Early news example: [Li et al., 2010]
Example:
What to show first?
?
...
Recommendation as
Contextual Bandit
● Environment: Netflix homepage
● Context: Member
● Arm: Display video at top of page
● Policy: Selects a video to recommend
● Reward: Member plays and enjoys video
Video Selector
▶
?
Winner
Bandit
Features
Model 1
Model 2
Model 3
Model 4
Member
(context)
Video
(arm)
Probability of
enjoyment
(Predicted reward)
Causality & Bandits [Dimakopoulou et al., 2021]
● Data collected from bandits is not IID
○ Bandits collect data adaptively
○ Initial noise may mean choosing an arm less
often, which can keep its sample mean low
● Inverse Propensity Weighting? High variance
○ Take inspiration from Doubly Robust
estimators
● Doubly Adaptive Thompson Sampling (DATS)
○ Thompson Sampling using the distribution of
the Adaptive Doubly Robust estimator in
place of the posterior
○ DATS performs better in practice and
matches TS regret bound
● Designing good exploration is an art
○ Especially to support future algorithm innovation
○ Challenging to do member-level A/B tests comparing
fully on-policy bandits at high scale
● Bandits over large action spaces: rankings and slates
● Layers of bandits that influence each other
● Handling delayed rewards
Challenges with bandits in the real world
Going Long-Term
● Want to maximize long-term member joy
● Involves many member visits, recommendation actions and delayed reward
● … sounds like Reinforcement Learning
Within a page
RL to optimize a
ranking or slate
How long?
Within a session
RL to optimize
multiple interactions
in a session
Across sessions
RL to optimize
interactions across
multiple sessions
Building simulators for evaluating recommenders
Page-level
Whole system (Accordion)
[McInerney et al., 2021]
Ranking
● Embeddings for actions: List-wise [Zhao et al., 2017] or Page-wise
recommendation [Zhao et al. 2018] based on [Dulac-Arnold et al., 2016]
● Adversarial model for user simulator: GAN-like model [Chen et al., 2019]
● Policy Gradient: Candidate generator using REINFORCE and TPRO [Chen
et al., 2019],
● Multi-task: Additional model head or Actor-Critic [Xin et al., 2020],
Auxiliary tasks for REINFORCE [Chen et al., 2021]
● Handling Diversity [Hansen et al., 2021], Slates [Ie et al., 2020], &
Multiple Recommenders [Zhao et. al, 2020]
● ...
Many potential directions
Trend 4: Objectives
● We want to optimize long-term member joy
● While accounting for:
○ Avoiding “trust busters”
○ Coldstarting
○ Fairness
○ Findability
○ ...
What is your recommender trying to optimize?
Layers of Metrics
Training
Objective
Offline Metric Online Metric Goal
Layers of Metrics
RMSE
NDCG on
historical data
Member
Engagement in
A/B test
Joy
Example case: Misaligned Metrics
Training
Objective
Offline Metric Online Metric Goal
Our recommendations can only be as good as the
metrics we measure it on
Recap [More et al., 2019]
● Bandit replay-style metrics can
have high variance due to low
number of matches with large
action spaces
● Use a ranking approach: Good to
rank high reward arms near top, low
reward arms near bottom
● Nuanced metrics:
○ Differences between what you want and what you can encapsulate in a
metric
○ Where does enjoyment come from? How does that vary by person?
○ How do you measure that at scale?
● What about effects beyond typical A/B time horizon?
● Incorporating fairness
○ Calibration to distribution of user tastes [Steck, 2018]
○ Item cold-start [Zhu et. al, 2021]
● Beyond algorithms: Ensuring a positive impact on society
Challenges in objectives
Conclusion
1. Deep Learning
2. Causality
3. Bandits & Reinforcement Learning
4. Objectives
A few recent trends in personalization
Sound interesting?Join us
research.netflix.com/jobs
Thank you
Anuj Shah
@badshah79
https://www.linkedin.com/in/foranuj/

Weitere ähnliche Inhalte

Was ist angesagt?

Learning a Personalized Homepage
Learning a Personalized HomepageLearning a Personalized Homepage
Learning a Personalized HomepageJustin Basilico
 
Déjà Vu: The Importance of Time and Causality in Recommender Systems
Déjà Vu: The Importance of Time and Causality in Recommender SystemsDéjà Vu: The Importance of Time and Causality in Recommender Systems
Déjà Vu: The Importance of Time and Causality in Recommender SystemsJustin Basilico
 
Sequential Decision Making in Recommendations
Sequential Decision Making in RecommendationsSequential Decision Making in Recommendations
Sequential Decision Making in RecommendationsJaya Kawale
 
Deeper Things: How Netflix Leverages Deep Learning in Recommendations and Se...
 Deeper Things: How Netflix Leverages Deep Learning in Recommendations and Se... Deeper Things: How Netflix Leverages Deep Learning in Recommendations and Se...
Deeper Things: How Netflix Leverages Deep Learning in Recommendations and Se...Sudeep Das, Ph.D.
 
Past, present, and future of Recommender Systems: an industry perspective
Past, present, and future of Recommender Systems: an industry perspectivePast, present, and future of Recommender Systems: an industry perspective
Past, present, and future of Recommender Systems: an industry perspectiveXavier Amatriain
 
Tutorial on Deep Learning in Recommender System, Lars summer school 2019
Tutorial on Deep Learning in Recommender System, Lars summer school 2019Tutorial on Deep Learning in Recommender System, Lars summer school 2019
Tutorial on Deep Learning in Recommender System, Lars summer school 2019Anoop Deoras
 
Missing values in recommender models
Missing values in recommender modelsMissing values in recommender models
Missing values in recommender modelsParmeshwar Khurd
 
Artwork Personalization at Netflix
Artwork Personalization at NetflixArtwork Personalization at Netflix
Artwork Personalization at NetflixJustin Basilico
 
Calibrated Recommendations
Calibrated RecommendationsCalibrated Recommendations
Calibrated RecommendationsHarald Steck
 
Making Netflix Machine Learning Algorithms Reliable
Making Netflix Machine Learning Algorithms ReliableMaking Netflix Machine Learning Algorithms Reliable
Making Netflix Machine Learning Algorithms ReliableJustin Basilico
 
Crafting Recommenders: the Shallow and the Deep of it!
Crafting Recommenders: the Shallow and the Deep of it! Crafting Recommenders: the Shallow and the Deep of it!
Crafting Recommenders: the Shallow and the Deep of it! Sudeep Das, Ph.D.
 
Shallow and Deep Latent Models for Recommender System
Shallow and Deep Latent Models for Recommender SystemShallow and Deep Latent Models for Recommender System
Shallow and Deep Latent Models for Recommender SystemAnoop Deoras
 
Netflix Recommendations Feature Engineering with Time Travel
Netflix Recommendations Feature Engineering with Time TravelNetflix Recommendations Feature Engineering with Time Travel
Netflix Recommendations Feature Engineering with Time TravelFaisal Siddiqi
 
Deep Learning for Recommender Systems
Deep Learning for Recommender SystemsDeep Learning for Recommender Systems
Deep Learning for Recommender SystemsYves Raimond
 
RecSys 2020 A Human Perspective on Algorithmic Similarity Schendel 9-2020
RecSys 2020 A Human Perspective on Algorithmic Similarity Schendel 9-2020RecSys 2020 A Human Perspective on Algorithmic Similarity Schendel 9-2020
RecSys 2020 A Human Perspective on Algorithmic Similarity Schendel 9-2020Zachary Schendel
 
A Multi-Armed Bandit Framework For Recommendations at Netflix
A Multi-Armed Bandit Framework For Recommendations at NetflixA Multi-Armed Bandit Framework For Recommendations at Netflix
A Multi-Armed Bandit Framework For Recommendations at NetflixJaya Kawale
 
Deep Learning for Recommender Systems
Deep Learning for Recommender SystemsDeep Learning for Recommender Systems
Deep Learning for Recommender SystemsJustin Basilico
 
Personalized Page Generation for Browsing Recommendations
Personalized Page Generation for Browsing RecommendationsPersonalized Page Generation for Browsing Recommendations
Personalized Page Generation for Browsing RecommendationsJustin Basilico
 
Lessons Learned from Building Machine Learning Software at Netflix
Lessons Learned from Building Machine Learning Software at NetflixLessons Learned from Building Machine Learning Software at Netflix
Lessons Learned from Building Machine Learning Software at NetflixJustin Basilico
 

Was ist angesagt? (20)

Learning a Personalized Homepage
Learning a Personalized HomepageLearning a Personalized Homepage
Learning a Personalized Homepage
 
Déjà Vu: The Importance of Time and Causality in Recommender Systems
Déjà Vu: The Importance of Time and Causality in Recommender SystemsDéjà Vu: The Importance of Time and Causality in Recommender Systems
Déjà Vu: The Importance of Time and Causality in Recommender Systems
 
Sequential Decision Making in Recommendations
Sequential Decision Making in RecommendationsSequential Decision Making in Recommendations
Sequential Decision Making in Recommendations
 
Deeper Things: How Netflix Leverages Deep Learning in Recommendations and Se...
 Deeper Things: How Netflix Leverages Deep Learning in Recommendations and Se... Deeper Things: How Netflix Leverages Deep Learning in Recommendations and Se...
Deeper Things: How Netflix Leverages Deep Learning in Recommendations and Se...
 
Past, present, and future of Recommender Systems: an industry perspective
Past, present, and future of Recommender Systems: an industry perspectivePast, present, and future of Recommender Systems: an industry perspective
Past, present, and future of Recommender Systems: an industry perspective
 
Tutorial on Deep Learning in Recommender System, Lars summer school 2019
Tutorial on Deep Learning in Recommender System, Lars summer school 2019Tutorial on Deep Learning in Recommender System, Lars summer school 2019
Tutorial on Deep Learning in Recommender System, Lars summer school 2019
 
Missing values in recommender models
Missing values in recommender modelsMissing values in recommender models
Missing values in recommender models
 
Artwork Personalization at Netflix
Artwork Personalization at NetflixArtwork Personalization at Netflix
Artwork Personalization at Netflix
 
Calibrated Recommendations
Calibrated RecommendationsCalibrated Recommendations
Calibrated Recommendations
 
Making Netflix Machine Learning Algorithms Reliable
Making Netflix Machine Learning Algorithms ReliableMaking Netflix Machine Learning Algorithms Reliable
Making Netflix Machine Learning Algorithms Reliable
 
Crafting Recommenders: the Shallow and the Deep of it!
Crafting Recommenders: the Shallow and the Deep of it! Crafting Recommenders: the Shallow and the Deep of it!
Crafting Recommenders: the Shallow and the Deep of it!
 
Shallow and Deep Latent Models for Recommender System
Shallow and Deep Latent Models for Recommender SystemShallow and Deep Latent Models for Recommender System
Shallow and Deep Latent Models for Recommender System
 
Netflix Recommendations Feature Engineering with Time Travel
Netflix Recommendations Feature Engineering with Time TravelNetflix Recommendations Feature Engineering with Time Travel
Netflix Recommendations Feature Engineering with Time Travel
 
Deep Learning for Recommender Systems
Deep Learning for Recommender SystemsDeep Learning for Recommender Systems
Deep Learning for Recommender Systems
 
RecSys 2020 A Human Perspective on Algorithmic Similarity Schendel 9-2020
RecSys 2020 A Human Perspective on Algorithmic Similarity Schendel 9-2020RecSys 2020 A Human Perspective on Algorithmic Similarity Schendel 9-2020
RecSys 2020 A Human Perspective on Algorithmic Similarity Schendel 9-2020
 
A Multi-Armed Bandit Framework For Recommendations at Netflix
A Multi-Armed Bandit Framework For Recommendations at NetflixA Multi-Armed Bandit Framework For Recommendations at Netflix
A Multi-Armed Bandit Framework For Recommendations at Netflix
 
Learning to Personalize
Learning to PersonalizeLearning to Personalize
Learning to Personalize
 
Deep Learning for Recommender Systems
Deep Learning for Recommender SystemsDeep Learning for Recommender Systems
Deep Learning for Recommender Systems
 
Personalized Page Generation for Browsing Recommendations
Personalized Page Generation for Browsing RecommendationsPersonalized Page Generation for Browsing Recommendations
Personalized Page Generation for Browsing Recommendations
 
Lessons Learned from Building Machine Learning Software at Netflix
Lessons Learned from Building Machine Learning Software at NetflixLessons Learned from Building Machine Learning Software at Netflix
Lessons Learned from Building Machine Learning Software at Netflix
 

Ähnlich wie Recent Trends in Personalization at Netflix

Recommender Systems In Industry
Recommender Systems In IndustryRecommender Systems In Industry
Recommender Systems In IndustryXavier Amatriain
 
Big & Personal: the data and the models behind Netflix recommendations by Xa...
 Big & Personal: the data and the models behind Netflix recommendations by Xa... Big & Personal: the data and the models behind Netflix recommendations by Xa...
Big & Personal: the data and the models behind Netflix recommendations by Xa...BigMine
 
Qualitative Research vs Quantitative Research - a QuestionPro Academic Webinar
Qualitative Research vs Quantitative Research - a QuestionPro Academic WebinarQualitative Research vs Quantitative Research - a QuestionPro Academic Webinar
Qualitative Research vs Quantitative Research - a QuestionPro Academic WebinarQuestionPro
 
Pivotal Tracker - Research Findings
Pivotal Tracker - Research FindingsPivotal Tracker - Research Findings
Pivotal Tracker - Research FindingsPaulina Galindo
 
Requirements Engineering for the Humanities
Requirements Engineering for the HumanitiesRequirements Engineering for the Humanities
Requirements Engineering for the HumanitiesShawn Day
 
User Experience Research: Deriving Insights for Customer Development
User Experience Research: Deriving Insights for Customer DevelopmentUser Experience Research: Deriving Insights for Customer Development
User Experience Research: Deriving Insights for Customer DevelopmentNoreen Whysel
 
KELOMPOK 4 -MSM 23.pptx
KELOMPOK 4 -MSM 23.pptxKELOMPOK 4 -MSM 23.pptx
KELOMPOK 4 -MSM 23.pptxWannabetheone
 
Understanding User Experience Workshop - Interlink Conference 2012
Understanding User Experience Workshop - Interlink Conference 2012Understanding User Experience Workshop - Interlink Conference 2012
Understanding User Experience Workshop - Interlink Conference 2012Lynne Polischuik
 
Intranet and collaboration - developing scenarios to define prioritisation an...
Intranet and collaboration - developing scenarios to define prioritisation an...Intranet and collaboration - developing scenarios to define prioritisation an...
Intranet and collaboration - developing scenarios to define prioritisation an...GabrieleSani3
 
Atd advanced topicsworkshop
Atd advanced topicsworkshopAtd advanced topicsworkshop
Atd advanced topicsworkshoplisacrispin
 
The Hive Think Tank: Machine Learning at Pinterest by Jure Leskovec
The Hive Think Tank: Machine Learning at Pinterest by Jure LeskovecThe Hive Think Tank: Machine Learning at Pinterest by Jure Leskovec
The Hive Think Tank: Machine Learning at Pinterest by Jure LeskovecThe Hive
 
Recommendation at Netflix Scale
Recommendation at Netflix ScaleRecommendation at Netflix Scale
Recommendation at Netflix ScaleJustin Basilico
 
Creating Data Driven Customer Profiles - Dawn of the Data Age Lecture Series
Creating Data Driven Customer Profiles - Dawn of the Data Age Lecture SeriesCreating Data Driven Customer Profiles - Dawn of the Data Age Lecture Series
Creating Data Driven Customer Profiles - Dawn of the Data Age Lecture SeriesLuciano Pesci, PhD
 
Machine Learning Product Managers Meetup Event
Machine Learning Product Managers Meetup EventMachine Learning Product Managers Meetup Event
Machine Learning Product Managers Meetup EventBenjamin Schulte
 
Qcon SF 2013 - Machine Learning & Recommender Systems @ Netflix Scale
Qcon SF 2013 - Machine Learning & Recommender Systems @ Netflix ScaleQcon SF 2013 - Machine Learning & Recommender Systems @ Netflix Scale
Qcon SF 2013 - Machine Learning & Recommender Systems @ Netflix ScaleXavier Amatriain
 
Recsys 2016 tutorial: Lessons learned from building real-life recommender sys...
Recsys 2016 tutorial: Lessons learned from building real-life recommender sys...Recsys 2016 tutorial: Lessons learned from building real-life recommender sys...
Recsys 2016 tutorial: Lessons learned from building real-life recommender sys...Xavier Amatriain
 
Recommandation systems -
Recommandation systems - Recommandation systems -
Recommandation systems - Yousef Fadila
 
2022-October In-person Meetup-Barcelona Admins Group.pdf
2022-October In-person Meetup-Barcelona Admins Group.pdf2022-October In-person Meetup-Barcelona Admins Group.pdf
2022-October In-person Meetup-Barcelona Admins Group.pdfanimuscrm
 
Machine Learning - Startup weekend UCSB 2018
Machine Learning - Startup weekend UCSB 2018Machine Learning - Startup weekend UCSB 2018
Machine Learning - Startup weekend UCSB 2018Raul Eulogio
 
Final project presentation Software Project 1
Final project presentation Software Project 1Final project presentation Software Project 1
Final project presentation Software Project 1MohtashimMasuk
 

Ähnlich wie Recent Trends in Personalization at Netflix (20)

Recommender Systems In Industry
Recommender Systems In IndustryRecommender Systems In Industry
Recommender Systems In Industry
 
Big & Personal: the data and the models behind Netflix recommendations by Xa...
 Big & Personal: the data and the models behind Netflix recommendations by Xa... Big & Personal: the data and the models behind Netflix recommendations by Xa...
Big & Personal: the data and the models behind Netflix recommendations by Xa...
 
Qualitative Research vs Quantitative Research - a QuestionPro Academic Webinar
Qualitative Research vs Quantitative Research - a QuestionPro Academic WebinarQualitative Research vs Quantitative Research - a QuestionPro Academic Webinar
Qualitative Research vs Quantitative Research - a QuestionPro Academic Webinar
 
Pivotal Tracker - Research Findings
Pivotal Tracker - Research FindingsPivotal Tracker - Research Findings
Pivotal Tracker - Research Findings
 
Requirements Engineering for the Humanities
Requirements Engineering for the HumanitiesRequirements Engineering for the Humanities
Requirements Engineering for the Humanities
 
User Experience Research: Deriving Insights for Customer Development
User Experience Research: Deriving Insights for Customer DevelopmentUser Experience Research: Deriving Insights for Customer Development
User Experience Research: Deriving Insights for Customer Development
 
KELOMPOK 4 -MSM 23.pptx
KELOMPOK 4 -MSM 23.pptxKELOMPOK 4 -MSM 23.pptx
KELOMPOK 4 -MSM 23.pptx
 
Understanding User Experience Workshop - Interlink Conference 2012
Understanding User Experience Workshop - Interlink Conference 2012Understanding User Experience Workshop - Interlink Conference 2012
Understanding User Experience Workshop - Interlink Conference 2012
 
Intranet and collaboration - developing scenarios to define prioritisation an...
Intranet and collaboration - developing scenarios to define prioritisation an...Intranet and collaboration - developing scenarios to define prioritisation an...
Intranet and collaboration - developing scenarios to define prioritisation an...
 
Atd advanced topicsworkshop
Atd advanced topicsworkshopAtd advanced topicsworkshop
Atd advanced topicsworkshop
 
The Hive Think Tank: Machine Learning at Pinterest by Jure Leskovec
The Hive Think Tank: Machine Learning at Pinterest by Jure LeskovecThe Hive Think Tank: Machine Learning at Pinterest by Jure Leskovec
The Hive Think Tank: Machine Learning at Pinterest by Jure Leskovec
 
Recommendation at Netflix Scale
Recommendation at Netflix ScaleRecommendation at Netflix Scale
Recommendation at Netflix Scale
 
Creating Data Driven Customer Profiles - Dawn of the Data Age Lecture Series
Creating Data Driven Customer Profiles - Dawn of the Data Age Lecture SeriesCreating Data Driven Customer Profiles - Dawn of the Data Age Lecture Series
Creating Data Driven Customer Profiles - Dawn of the Data Age Lecture Series
 
Machine Learning Product Managers Meetup Event
Machine Learning Product Managers Meetup EventMachine Learning Product Managers Meetup Event
Machine Learning Product Managers Meetup Event
 
Qcon SF 2013 - Machine Learning & Recommender Systems @ Netflix Scale
Qcon SF 2013 - Machine Learning & Recommender Systems @ Netflix ScaleQcon SF 2013 - Machine Learning & Recommender Systems @ Netflix Scale
Qcon SF 2013 - Machine Learning & Recommender Systems @ Netflix Scale
 
Recsys 2016 tutorial: Lessons learned from building real-life recommender sys...
Recsys 2016 tutorial: Lessons learned from building real-life recommender sys...Recsys 2016 tutorial: Lessons learned from building real-life recommender sys...
Recsys 2016 tutorial: Lessons learned from building real-life recommender sys...
 
Recommandation systems -
Recommandation systems - Recommandation systems -
Recommandation systems -
 
2022-October In-person Meetup-Barcelona Admins Group.pdf
2022-October In-person Meetup-Barcelona Admins Group.pdf2022-October In-person Meetup-Barcelona Admins Group.pdf
2022-October In-person Meetup-Barcelona Admins Group.pdf
 
Machine Learning - Startup weekend UCSB 2018
Machine Learning - Startup weekend UCSB 2018Machine Learning - Startup weekend UCSB 2018
Machine Learning - Startup weekend UCSB 2018
 
Final project presentation Software Project 1
Final project presentation Software Project 1Final project presentation Software Project 1
Final project presentation Software Project 1
 

Mehr von Förderverein Technische Fakultät

The Digital Transformation of Education: A Hyper-Disruptive Era through Block...
The Digital Transformation of Education: A Hyper-Disruptive Era through Block...The Digital Transformation of Education: A Hyper-Disruptive Era through Block...
The Digital Transformation of Education: A Hyper-Disruptive Era through Block...Förderverein Technische Fakultät
 
Engineering Serverless Workflow Applications in Federated FaaS.pdf
Engineering Serverless Workflow Applications in Federated FaaS.pdfEngineering Serverless Workflow Applications in Federated FaaS.pdf
Engineering Serverless Workflow Applications in Federated FaaS.pdfFörderverein Technische Fakultät
 
The Role of Machine Learning in Fluid Network Control and Data Planes.pdf
The Role of Machine Learning in Fluid Network Control and Data Planes.pdfThe Role of Machine Learning in Fluid Network Control and Data Planes.pdf
The Role of Machine Learning in Fluid Network Control and Data Planes.pdfFörderverein Technische Fakultät
 
Nonequilibrium Network Dynamics_Inference, Fluctuation-Respones & Tipping Poi...
Nonequilibrium Network Dynamics_Inference, Fluctuation-Respones & Tipping Poi...Nonequilibrium Network Dynamics_Inference, Fluctuation-Respones & Tipping Poi...
Nonequilibrium Network Dynamics_Inference, Fluctuation-Respones & Tipping Poi...Förderverein Technische Fakultät
 
East-west oriented photovoltaic power systems: model, benefits and technical ...
East-west oriented photovoltaic power systems: model, benefits and technical ...East-west oriented photovoltaic power systems: model, benefits and technical ...
East-west oriented photovoltaic power systems: model, benefits and technical ...Förderverein Technische Fakultät
 
Advances in Visual Quality Restoration with Generative Adversarial Networks
Advances in Visual Quality Restoration with Generative Adversarial NetworksAdvances in Visual Quality Restoration with Generative Adversarial Networks
Advances in Visual Quality Restoration with Generative Adversarial NetworksFörderverein Technische Fakultät
 
Industriepraktikum_ Unterstützung bei Projekten in der Automatisierung.pdf
Industriepraktikum_ Unterstützung bei Projekten in der Automatisierung.pdfIndustriepraktikum_ Unterstützung bei Projekten in der Automatisierung.pdf
Industriepraktikum_ Unterstützung bei Projekten in der Automatisierung.pdfFörderverein Technische Fakultät
 

Mehr von Förderverein Technische Fakultät (20)

Supervisory control of business processes
Supervisory control of business processesSupervisory control of business processes
Supervisory control of business processes
 
The Digital Transformation of Education: A Hyper-Disruptive Era through Block...
The Digital Transformation of Education: A Hyper-Disruptive Era through Block...The Digital Transformation of Education: A Hyper-Disruptive Era through Block...
The Digital Transformation of Education: A Hyper-Disruptive Era through Block...
 
A Game of Chess is Like a Swordfight.pdf
A Game of Chess is Like a Swordfight.pdfA Game of Chess is Like a Swordfight.pdf
A Game of Chess is Like a Swordfight.pdf
 
From Mind to Meta.pdf
From Mind to Meta.pdfFrom Mind to Meta.pdf
From Mind to Meta.pdf
 
Miniatures Design for Tabletop Games.pdf
Miniatures Design for Tabletop Games.pdfMiniatures Design for Tabletop Games.pdf
Miniatures Design for Tabletop Games.pdf
 
Distributed Systems in the Post-Moore Era.pptx
Distributed Systems in the Post-Moore Era.pptxDistributed Systems in the Post-Moore Era.pptx
Distributed Systems in the Post-Moore Era.pptx
 
Don't Treat the Symptom, Find the Cause!.pptx
Don't Treat the Symptom, Find the Cause!.pptxDon't Treat the Symptom, Find the Cause!.pptx
Don't Treat the Symptom, Find the Cause!.pptx
 
Engineering Serverless Workflow Applications in Federated FaaS.pdf
Engineering Serverless Workflow Applications in Federated FaaS.pdfEngineering Serverless Workflow Applications in Federated FaaS.pdf
Engineering Serverless Workflow Applications in Federated FaaS.pdf
 
The Role of Machine Learning in Fluid Network Control and Data Planes.pdf
The Role of Machine Learning in Fluid Network Control and Data Planes.pdfThe Role of Machine Learning in Fluid Network Control and Data Planes.pdf
The Role of Machine Learning in Fluid Network Control and Data Planes.pdf
 
Nonequilibrium Network Dynamics_Inference, Fluctuation-Respones & Tipping Poi...
Nonequilibrium Network Dynamics_Inference, Fluctuation-Respones & Tipping Poi...Nonequilibrium Network Dynamics_Inference, Fluctuation-Respones & Tipping Poi...
Nonequilibrium Network Dynamics_Inference, Fluctuation-Respones & Tipping Poi...
 
Towards a data driven identification of teaching patterns.pdf
Towards a data driven identification of teaching patterns.pdfTowards a data driven identification of teaching patterns.pdf
Towards a data driven identification of teaching patterns.pdf
 
Förderverein Technische Fakultät.pptx
Förderverein Technische Fakultät.pptxFörderverein Technische Fakultät.pptx
Förderverein Technische Fakultät.pptx
 
The Computing Continuum.pdf
The Computing Continuum.pdfThe Computing Continuum.pdf
The Computing Continuum.pdf
 
East-west oriented photovoltaic power systems: model, benefits and technical ...
East-west oriented photovoltaic power systems: model, benefits and technical ...East-west oriented photovoltaic power systems: model, benefits and technical ...
East-west oriented photovoltaic power systems: model, benefits and technical ...
 
Machine Learning in Finance via Randomization
Machine Learning in Finance via RandomizationMachine Learning in Finance via Randomization
Machine Learning in Finance via Randomization
 
IT does not stop
IT does not stopIT does not stop
IT does not stop
 
Advances in Visual Quality Restoration with Generative Adversarial Networks
Advances in Visual Quality Restoration with Generative Adversarial NetworksAdvances in Visual Quality Restoration with Generative Adversarial Networks
Advances in Visual Quality Restoration with Generative Adversarial Networks
 
Industriepraktikum_ Unterstützung bei Projekten in der Automatisierung.pdf
Industriepraktikum_ Unterstützung bei Projekten in der Automatisierung.pdfIndustriepraktikum_ Unterstützung bei Projekten in der Automatisierung.pdf
Industriepraktikum_ Unterstützung bei Projekten in der Automatisierung.pdf
 
Introduction to 5G from radio perspective
Introduction to 5G from radio perspectiveIntroduction to 5G from radio perspective
Introduction to 5G from radio perspective
 
Förderverein Technische Fakultät
Förderverein Technische Fakultät Förderverein Technische Fakultät
Förderverein Technische Fakultät
 

Kürzlich hochgeladen

Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)Allon Mureinik
 
Automating Google Workspace (GWS) & more with Apps Script
Automating Google Workspace (GWS) & more with Apps ScriptAutomating Google Workspace (GWS) & more with Apps Script
Automating Google Workspace (GWS) & more with Apps Scriptwesley chun
 
[2024]Digital Global Overview Report 2024 Meltwater.pdf
[2024]Digital Global Overview Report 2024 Meltwater.pdf[2024]Digital Global Overview Report 2024 Meltwater.pdf
[2024]Digital Global Overview Report 2024 Meltwater.pdfhans926745
 
08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking Men08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking MenDelhi Call girls
 
Exploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone ProcessorsExploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone Processorsdebabhi2
 
Salesforce Community Group Quito, Salesforce 101
Salesforce Community Group Quito, Salesforce 101Salesforce Community Group Quito, Salesforce 101
Salesforce Community Group Quito, Salesforce 101Paola De la Torre
 
A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024Results
 
Unblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesUnblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesSinan KOZAK
 
Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024The Digital Insurer
 
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...Drew Madelung
 
Histor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slideHistor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slidevu2urc
 
GenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day PresentationGenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day PresentationMichael W. Hawkins
 
08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking Men08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking MenDelhi Call girls
 
2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...Martijn de Jong
 
Presentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreterPresentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreternaman860154
 
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...Neo4j
 
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptxHampshireHUG
 
A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)Gabriella Davis
 
CNv6 Instructor Chapter 6 Quality of Service
CNv6 Instructor Chapter 6 Quality of ServiceCNv6 Instructor Chapter 6 Quality of Service
CNv6 Instructor Chapter 6 Quality of Servicegiselly40
 
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdfThe Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdfEnterprise Knowledge
 

Kürzlich hochgeladen (20)

Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)
 
Automating Google Workspace (GWS) & more with Apps Script
Automating Google Workspace (GWS) & more with Apps ScriptAutomating Google Workspace (GWS) & more with Apps Script
Automating Google Workspace (GWS) & more with Apps Script
 
[2024]Digital Global Overview Report 2024 Meltwater.pdf
[2024]Digital Global Overview Report 2024 Meltwater.pdf[2024]Digital Global Overview Report 2024 Meltwater.pdf
[2024]Digital Global Overview Report 2024 Meltwater.pdf
 
08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking Men08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking Men
 
Exploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone ProcessorsExploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone Processors
 
Salesforce Community Group Quito, Salesforce 101
Salesforce Community Group Quito, Salesforce 101Salesforce Community Group Quito, Salesforce 101
Salesforce Community Group Quito, Salesforce 101
 
A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024
 
Unblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesUnblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen Frames
 
Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024
 
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
 
Histor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slideHistor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slide
 
GenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day PresentationGenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day Presentation
 
08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking Men08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking Men
 
2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...
 
Presentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreterPresentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreter
 
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
 
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
 
A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)
 
CNv6 Instructor Chapter 6 Quality of Service
CNv6 Instructor Chapter 6 Quality of ServiceCNv6 Instructor Chapter 6 Quality of Service
CNv6 Instructor Chapter 6 Quality of Service
 
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdfThe Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
 

Recent Trends in Personalization at Netflix

  • 1. Recent Trends in Personalization at Netflix Anuj Shah @badshah79 https://www.linkedin.com/in/foranuj/
  • 2. Why do we personalize?
  • 3. Help members find entertainment to watch and enjoy to maximize member satisfaction and retention
  • 5. What do we personalize?
  • 6. Ordering of videos is personalized From how we rank Ranking
  • 7. Selection and placement of rows is personalized ... to how we construct a page Rows
  • 8. ... to how we respond to queries Search query & result recommendation
  • 9. ... to how we cover different needs Personalized instant choices
  • 10. ... to how we reach out Message personalization
  • 11. Everything is a recommendation!
  • 13. ○ Every person is unique with a variety of interests … and sometimes multiple people use the same profile ○ Help people find what they want when they’re not sure what they want ○ Non-stationary, context-dependent, mood-dependent, ... ○ Large datasets but small data per member … and potentially biased by the output of your system ○ Cold-start problems on all sides ○ More than just accuracy: diversity, novelty, freshness, fairness, ... ○ ... No, personalization is hard!
  • 14. So how are we going to solve this?
  • 15. Some recent avenues in approaching these challenges: 1. Deep Learning 2. Causality 3. Bandits & Reinforcement Learning 4. Objectives Trending Now
  • 16. Trend 1: Deep Learning for Recommendations
  • 17. ~2012 ~2017 Deep Learning becomes popular in Machine Learning Deep Learning becomes popular in Recommender Systems What took so long? ~2019 Traditional methods do as well or better than Deep Learning for Recommender Systems … Wait, what? Timeline
  • 18. Traditional Recommendations Collaborative Filtering: Recommend items that similar users have chosen 0 1 0 1 0 0 0 1 1 0 1 0 0 1 1 0 1 0 0 0 0 0 0 0 1 Users Items
  • 21. U A (deeper) feed-forward view V Mean squared loss ?
  • 22. … isn’t always the best U V Mean squared loss ? Also see [Dacrema et al., 2019], [Rendle et. al, 2019], [Rendle et. al, 2021]. Make sure you tune your baselines.
  • 23. Understanding the relationships From our forthcoming AI Magazine article “Deep Learning for Recommender Systems: A Netflix Case-Study”
  • 24. X R EASE: Embarrassingly Shallow Auto-Encoders [Steck, 2019] ● Super efficient model to train in a collaborative filtering setting inspired by SLIM ● Learn item-by-item matrix X such that R.X is close to R and diag(X) is 0 ○ Avoids trivial solution of identity ● Closed-form solution ● More on that: auto-encoders that don’t overfit towards identity ≈ R 0 0
  • 25. Modern Recommender Systems 0 1 0 1 0 0 0 1 1 0 1 0 0 1 1 0 1 0 0 0 0 0 0 0 1 Users Items
  • 26. Contextual event data Modern Recommender Data Model Interactions Impressions Item data Profile settings +
  • 27. Contextual sequence data 2017-12-10 15:40:22 2017-12-23 19:32:10 2017-12-24 12:05:53 2017-12-27 22:40:22 2017-12-29 19:39:36 2017-12-30 20:42:13 Context Item Sequence per member ? Time
  • 28. V Sequential Recommendation Network Softmax over items Avg / Stack / Sequence / Attention DNN / RNN / CNN / TNN Input interactions (X) (X) p(Y|X) 2018-12-23 19:32:10 2018-12-24 12:05:53 2019-01-02 15:40:22
  • 31. From Correlation to Causation ● Most recommendation algorithms are correlational ○ Some early recommendation algorithms literally computed correlations between users and items ● Did you watch a movie because we recommended it to you? Or because you liked it? Or both? ● If you had to watch a movie, would you like it? [Wang et al., 2020] p(Y|X) → p(Y|X, do(R)) (from http://www.tylervigen.com/spurious-correlations)
  • 32. Feedback loops Impression bias inflates plays Leads to inflated item popularity More plays More impressions Oscillations in distribution of genre recommendations Feedback loops can cause biases to be reinforced by the recommendation system! [Chaney et al., 2018]: simulations showing that this can reduce the usefulness of the system They’re real:
  • 33. Lots of feedback loops...
  • 35. Closed Loop Training Data Watches Model Recs Danger Zone Search Training Data Watches Model Recs Open Loop
  • 36. Closed Loop Training Data Watches Model Recs Danger Zone Search Training Data Watches Model Recs Open Loop
  • 37. V Propensity Correction Avg / Stack / Sequence / Attention DNN / RNN / CNN / TNN Input interactions (X) (X) 2018-12-23 19:32:10 2018-12-24 12:05:53 2019-01-02 15:40:22 Policy softmax Propensity softmax E.g. [Chen et al., 2019] p(Y|X, do(R)) p(R|X)
  • 38. Challenges in Causal Recommendations ● Handling unobserved confounders ● Coming up with the right causal graph ● High variance (especially propensity-based ones) ● Computational challenges (e.g. [Wong, 2020]) ● Off-policy evaluation ● When and how to introduce exploration
  • 39. Trend 3: Bandits & Reinforcement Learning in Recommendations
  • 40. Why contextual bandits for recommendations? ● Break feedback loops ● Want to explore to learn ● Uncertainty around member interests and new items ● Sparse and indirect feedback ● Changing trends ▶ Early news example: [Li et al., 2010]
  • 41. Example: What to show first? ? ...
  • 42. Recommendation as Contextual Bandit ● Environment: Netflix homepage ● Context: Member ● Arm: Display video at top of page ● Policy: Selects a video to recommend ● Reward: Member plays and enjoys video Video Selector ▶ ?
  • 43. Winner Bandit Features Model 1 Model 2 Model 3 Model 4 Member (context) Video (arm) Probability of enjoyment (Predicted reward)
  • 44. Causality & Bandits [Dimakopoulou et al., 2021] ● Data collected from bandits is not IID ○ Bandits collect data adaptively ○ Initial noise may mean choosing an arm less often, which can keep its sample mean low ● Inverse Propensity Weighting? High variance ○ Take inspiration from Doubly Robust estimators ● Doubly Adaptive Thompson Sampling (DATS) ○ Thompson Sampling using the distribution of the Adaptive Doubly Robust estimator in place of the posterior ○ DATS performs better in practice and matches TS regret bound
  • 45. ● Designing good exploration is an art ○ Especially to support future algorithm innovation ○ Challenging to do member-level A/B tests comparing fully on-policy bandits at high scale ● Bandits over large action spaces: rankings and slates ● Layers of bandits that influence each other ● Handling delayed rewards Challenges with bandits in the real world
  • 46. Going Long-Term ● Want to maximize long-term member joy ● Involves many member visits, recommendation actions and delayed reward ● … sounds like Reinforcement Learning
  • 47. Within a page RL to optimize a ranking or slate How long? Within a session RL to optimize multiple interactions in a session Across sessions RL to optimize interactions across multiple sessions
  • 48. Building simulators for evaluating recommenders Page-level Whole system (Accordion) [McInerney et al., 2021] Ranking
  • 49. ● Embeddings for actions: List-wise [Zhao et al., 2017] or Page-wise recommendation [Zhao et al. 2018] based on [Dulac-Arnold et al., 2016] ● Adversarial model for user simulator: GAN-like model [Chen et al., 2019] ● Policy Gradient: Candidate generator using REINFORCE and TPRO [Chen et al., 2019], ● Multi-task: Additional model head or Actor-Critic [Xin et al., 2020], Auxiliary tasks for REINFORCE [Chen et al., 2021] ● Handling Diversity [Hansen et al., 2021], Slates [Ie et al., 2020], & Multiple Recommenders [Zhao et. al, 2020] ● ... Many potential directions
  • 51. ● We want to optimize long-term member joy ● While accounting for: ○ Avoiding “trust busters” ○ Coldstarting ○ Fairness ○ Findability ○ ... What is your recommender trying to optimize?
  • 52. Layers of Metrics Training Objective Offline Metric Online Metric Goal
  • 53. Layers of Metrics RMSE NDCG on historical data Member Engagement in A/B test Joy Example case: Misaligned Metrics Training Objective Offline Metric Online Metric Goal
  • 54. Our recommendations can only be as good as the metrics we measure it on
  • 55. Recap [More et al., 2019] ● Bandit replay-style metrics can have high variance due to low number of matches with large action spaces ● Use a ranking approach: Good to rank high reward arms near top, low reward arms near bottom
  • 56. ● Nuanced metrics: ○ Differences between what you want and what you can encapsulate in a metric ○ Where does enjoyment come from? How does that vary by person? ○ How do you measure that at scale? ● What about effects beyond typical A/B time horizon? ● Incorporating fairness ○ Calibration to distribution of user tastes [Steck, 2018] ○ Item cold-start [Zhu et. al, 2021] ● Beyond algorithms: Ensuring a positive impact on society Challenges in objectives
  • 58. 1. Deep Learning 2. Causality 3. Bandits & Reinforcement Learning 4. Objectives A few recent trends in personalization