query based music recommendations via preference embedding
play

Query-based Music Recommendations via Preference Embedding - PowerPoint PPT Presentation

Jan 10, 2023 •4.02k likes •4.36k views

Query-based Music Recommendations via Preference Embedding Chih-Ming Chen, Ming-Feng Tsai, Yu-Ching Lin, Yi-Hsuan Yang. Institutes involved in this research work CLIP Lab, National Chengchi University MAC Lab,

  1. Query-based Music Recommendations via Preference Embedding Chih-­‑Ming ¡Chen, ¡Ming-­‑Feng ¡Tsai, ¡Yu-­‑Ching ¡Lin, ¡Yi-­‑Hsuan ¡Yang.

  2. Institutes involved in this research work CLIP Lab, National Chengchi University MAC Lab, CITI, Academia Sinica Machine Learning Team, KKBOX

  3. Query-based Music Recommendations

  4. Query-based Music Recommendations

  5. Query-based Music Recommendations

  6. Query-based Music Recommendations

  7. Query-based Music Recommendations

  8. Query-based Music Recommendations

  9. The Graph Embedding Models Vertices Relation Graph Latent Space

  10. The Graph Embedding Models How to build 
 How to learn 
 Vertices the Relation Graph? the representations?

  11. Construction of User Preference Network 112 16 8 - - - T1 User - 2 119 64 - - U1 T2 - - 32 - 109 5 Track - - - - 12 8 U2 T3 # of Listening / Rating / 
 Album Like / Dislike / … U3 T4 Artist T5 U4 T6

  12. Construction of User Preference Network 112 16 8 - - - T1 User - 2 119 64 - - U1 T2 - - 32 - 109 5 Track - - - - 12 8 U2 T3 Al1 Album U3 T4 Al2 112 24 - - - 121 64 - Artist T5 Al3 - 32 - 114 U4 - - - 20 Al4 T6

  13. Construction of User Preference Network Ar1 112 16 8 - - - T1 User Ar2 - 2 119 64 - - U1 T2 - - 32 - 109 5 Track Ar3 - - - - 12 8 U2 T3 Al1 Album U3 T4 Al2 112 24 - - 112 24 - - 121 64 - 121 64 - Artist T5 Al3 - 32 - 114 - 32 114 U4 - - - 20 - - 20 Al4 T6

  14. User Preference Network Edges: User Preference Ar1 T1 binary value / numerical value Ar2 U1 Bipartite Graph T2 Ar3 it’s similar to CF-based models U2 T3 Heterogeneous Graph Al1 T4 U3 it considers multiple entities Al2 T5 Al3 U4 T6 Al4

  15. User Preference Network Edges: User Preference Ar1 T1 binary value / numerical value Ar2 U1 Bipartite Graph T2 Ar3 it’s similar to CF-based models U2 T3 Heterogeneous Graph Al1 T4 U3 it considers multiple entities Al2 This is how we achieve the 
 Query-based recommendations T5 Al3 U4 T6 Al4

  16. Heterogeneous Preference Embedding (HPE) Pr( community( ) | ) Φ ( ) Ar1 U3 U3 T1 Ar2 U1 compress the info T2 Ar3 U2 T3 Al1 T4 U3 Al2 T5 Al3 U4 T6 Al4

  17. Heterogeneous Preference Embedding (HPE) Pr( community( ) | ) Φ ( ) Ar1 U3 U3 T1 Ar2 U1 Sample an Edge T2 Ar3 Pr( | ) U2 T3 U3 T3 Al1 T4 U3 Al2 T5 Al3 U4 T6 Al4

  18. Heterogeneous Preference Embedding (HPE) Pr( community( ) | ) Φ ( ) Ar1 U3 U3 T1 Ar2 U1 Sample an Edge Random Walk T2 Ar3 Pr( | ) Pr( | ) U2 T3 U3 U1 U3 T3 Al1 T4 Pr( | ) U3 Al2 Al1 U3 T5 Al3 U4 T6 Al4

  19. Heterogeneous Preference Embedding (HPE) Pr( community( ) | ) Φ ( ) Ar1 U3 U3 T1 Ar2 U1 Sample an Edge Random Walk T2 Ar3 Pr( | ) Pr( | ) U2 T3 U3 U1 U3 T3 Al1 T4 Pr( | ) U3 Al2 Al1 U3 T5 X X k Φ ( v i ) k 2 O = � w i,j log p ( v j | Φ ( v i )) + λ Al3 U4 i ( i,j ) 2 S + negative sampling T6 Al4

  20. Heterogeneous Preference Embedding (HPE) Pr( community( ) | ) Φ ( ) Ar1 U3 U3 T1 Ar2 U1 Sample an Edge Random Walk T2 Ar3 Pr( | ) Pr( | ) U2 T3 U3 U1 U3 T3 Al1 T4 Pr( | ) U3 Al2 Al1 U3 T5 X X k Φ ( v i ) k 2 O = � w i,j log p ( v j | Φ ( v i )) + λ Al3 U4 i ( i,j ) 2 S + negative sampling T6 Al4

  21. Performance of Preference Embedding HitRatio@10 lastfm-1k (window=5) KKBOX (window=5) MSD (original) d = 16 d = 32 d = 64 d = 64 d = 128 d = 256 d = 64 d = 128 d = 256 Popularity 2.66% 2.66% 2.66% 4.32% 4.32% 4.32% 0.92% 0.92% 0.92% MF 3.02% 3.93% 4.22% 7.11% 8.49% 8.93% 1.37% 1.79% 2.00% DeepWalk 3.18% 3.55% 3.54% 11.61% 12.55% 13.08% 1.71% 1.95% 1.95% LINE-2nd 3.44% 3.74% 4.10% 12.79% 13.47% 12.77% 1.62% 1.60% 1.14% Proposed PE 3.54 % * 4.22 % 4.51 % 12.95 % * 13.74 % * 14.20 % * 2.08 % * 2.15 % * 2.19 % mAP@10 lastfm-1k (window=5) KKBOX (window=5) MSD (original) d = 16 d = 32 d = 64 d = 64 d = 128 d = 256 d = 64 d = 128 d = 256 Popularity 3.27 % 3.27 % 3.27 % 5.03% 5.03% 5.03% 1.04% 1.04% 1.04% MF 1.87% 2.34% 2.60% 4.65% 5.85% 6.16% 1.88% 2.44% 2.81% DeepWalk 1.82% 2.10% 1.99% 8.73% 9.47% 10.01% 2.66% 2.70% 2.55% LINE-2nd 2.00% 2.10% 2.38% 9.95% 10.64% 10.09% 1.84% 1.60% 1.44% Proposed PE 2.08% 2.55% 2.71% 10.14 % 10.86 % * 11.31 % 2.86 % * 3.09 % * 3.12 %

  22. Performance of Preference Embedding HitRatio@10 lastfm-1k (window=5) KKBOX (window=5) MSD (original) d = 16 d = 32 d = 64 d = 64 d = 128 d = 256 d = 64 d = 128 d = 256 Popularity 2.66% 2.66% 2.66% 4.32% 4.32% 4.32% 0.92% 0.92% 0.92% MF 3.02% 3.93% 4.22% 7.11% 8.49% 8.93% 1.37% 1.79% 2.00% DeepWalk 3.18% 3.55% 3.54% 11.61% 12.55% 13.08% 1.71% 1.95% 1.95% LINE-2nd 3.44% 3.74% 4.10% 12.79% 13.47% 12.77% 1.62% 1.60% 1.14% Proposed PE 3.54 % * 4.22 % 4.51 % 12.95 % * 13.74 % * 14.20 % * 2.08 % * 2.15 % * 2.19 % mAP@10 lastfm-1k (window=5) KKBOX (window=5) MSD (original) d = 16 d = 32 d = 64 d = 64 d = 128 d = 256 d = 64 d = 128 d = 256 Popularity 3.27 % 3.27 % 3.27 % 5.03% 5.03% 5.03% 1.04% 1.04% 1.04% MF 1.87% 2.34% 2.60% 4.65% 5.85% 6.16% 1.88% 2.44% 2.81% DeepWalk 1.82% 2.10% 1.99% 8.73% 9.47% 10.01% 2.66% 2.70% 2.55% LINE-2nd 2.00% 2.10% 2.38% 9.95% 10.64% 10.09% 1.84% 1.60% 1.44% Proposed PE 2.08% 2.55% 2.71% 10.14 % 10.86 % * 11.31 % 2.86 % * 3.09 % * 3.12 %

  23. Performance of Preference Embedding HitRatio@10 lastfm-1k (window=5) KKBOX (window=5) MSD (original) d = 16 d = 32 d = 64 d = 64 d = 128 d = 256 d = 64 d = 128 d = 256 Popularity 2.66% 2.66% 2.66% 4.32% 4.32% 4.32% 0.92% 0.92% 0.92% MF 3.02% 3.93% 4.22% 7.11% 8.49% 8.93% 1.37% 1.79% 2.00% DeepWalk 3.18% 3.55% 3.54% 11.61% 12.55% 13.08% 1.71% 1.95% 1.95% LINE-2nd 3.44% 3.74% 4.10% 12.79% 13.47% 12.77% 1.62% 1.60% 1.14% Proposed PE 3.54 % * 4.22 % 4.51 % 12.95 % * 13.74 % * 14.20 % * 2.08 % * 2.15 % * 2.19 % mAP@10 lastfm-1k (window=5) KKBOX (window=5) MSD (original) d = 16 d = 32 d = 64 d = 64 d = 128 d = 256 d = 64 d = 128 d = 256 Popularity 3.27 % 3.27 % 3.27 % 5.03% 5.03% 5.03% 1.04% 1.04% 1.04% MF 1.87% 2.34% 2.60% 4.65% 5.85% 6.16% 1.88% 2.44% 2.81% DeepWalk 1.82% 2.10% 1.99% 8.73% 9.47% 10.01% 2.66% 2.70% 2.55% LINE-2nd 2.00% 2.10% 2.38% 9.95% 10.64% 10.09% 1.84% 1.60% 1.44% Proposed PE 2.08% 2.55% 2.71% 10.14 % 10.86 % * 11.31 % 2.86 % * 3.09 % * 3.12 %

  24. Performance of Preference Embedding HitRatio@10 lastfm-1k (window=5) KKBOX (window=5) MSD (original) d = 16 d = 32 d = 64 d = 64 d = 128 d = 256 d = 64 d = 128 d = 256 Popularity 2.66% 2.66% 2.66% 4.32% 4.32% 4.32% 0.92% 0.92% 0.92% MF 3.02% 3.93% 4.22% 7.11% 8.49% 8.93% 1.37% 1.79% 2.00% DeepWalk 3.18% 3.55% 3.54% 11.61% 12.55% 13.08% 1.71% 1.95% 1.95% LINE-2nd 3.44% 3.74% 4.10% 12.79% 13.47% 12.77% 1.62% 1.60% 1.14% Proposed PE 3.54 % * 4.22 % 4.51 % 12.95 % * 13.74 % * 14.20 % * 2.08 % * 2.15 % * 2.19 % mAP@10 lastfm-1k (window=5) KKBOX (window=5) MSD (original) d = 16 d = 32 d = 64 d = 64 d = 128 d = 256 d = 64 d = 128 d = 256 Popularity 3.27 % 3.27 % 3.27 % 5.03% 5.03% 5.03% 1.04% 1.04% 1.04% MF 1.87% 2.34% 2.60% 4.65% 5.85% 6.16% 1.88% 2.44% 2.81% DeepWalk 1.82% 2.10% 1.99% 8.73% 9.47% 10.01% 2.66% 2.70% 2.55% LINE-2nd 2.00% 2.10% 2.38% 9.95% 10.64% 10.09% 1.84% 1.60% 1.44% Proposed PE 2.08% 2.55% 2.71% 10.14 % 10.86 % * 11.31 % 2.86 % * 3.09 % * 3.12 %

  25. Performance of Preference Embedding HitRatio@10 lastfm-1k (window=5) KKBOX (window=5) MSD (original) d = 16 d = 32 d = 64 d = 64 d = 128 d = 256 d = 64 d = 128 d = 256 Popularity 2.66% 2.66% 2.66% 4.32% 4.32% 4.32% 0.92% 0.92% 0.92% MF 3.02% 3.93% 4.22% 7.11% 8.49% 8.93% 1.37% 1.79% 2.00% DeepWalk 3.18% 3.55% 3.54% 11.61% 12.55% 13.08% 1.71% 1.95% 1.95% LINE-2nd 3.44% 3.74% 4.10% 12.79% 13.47% 12.77% 1.62% 1.60% 1.14% Proposed PE 3.54 % * 4.22 % 4.51 % 12.95 % * 13.74 % * 14.20 % * 2.08 % * 2.15 % * 2.19 % mAP@10 lastfm-1k (window=5) KKBOX (window=5) MSD (original) d = 16 d = 32 d = 64 d = 64 d = 128 d = 256 d = 64 d = 128 d = 256 Popularity 3.27 % 3.27 % 3.27 % 5.03% 5.03% 5.03% 1.04% 1.04% 1.04% MF 1.87% 2.34% 2.60% 4.65% 5.85% 6.16% 1.88% 2.44% 2.81% DeepWalk 1.82% 2.10% 1.99% 8.73% 9.47% 10.01% 2.66% 2.70% 2.55% LINE-2nd 2.00% 2.10% 2.38% 9.95% 10.64% 10.09% 1.84% 1.60% 1.44% Proposed PE 2.08% 2.55% 2.71% 10.14 % 10.86 % * 11.31 % 2.86 % * 3.09 % * 3.12 %

  26. Extension Work

  27. Extension Work Multiple 
 Queries:

  28. Extension Work Multiple 
 Queries:

Recommend

Ranked Query Processing: Relevance a) Order-based Paradigm Preference Q Kevin Chen-Chuan

Ranked Query Processing: Relevance a) Order-based Paradigm Preference Q Kevin Chen-Chuan

Ranking Ordering according to the degree of some fuzzy notions: Similarity (or dissimilarity) Ranked Query Processing: Relevance a) Order-based Paradigm Preference Q Kevin Chen-Chuan Chang ranking 2 Query models for

295 views • 13 slides
3. Preference Learning Techniques a. Learning Utility Functions b. Learning Preference

3. Preference Learning Techniques a. Learning Utility Functions b. Learning Preference

AGENDA 1. Preference Learning Tasks 2. Performance Assessment and Loss Functions 3. Preference Learning Techniques a. Learning Utility Functions b. Learning Preference Relations c. Structured Output Prediction d. Model-Based Preference

583 views • 39 slides
Overview n Melody-Based Retrieval n Audio-Score Alignment n Music Fingerprinting 2 Carnegie

Overview n Melody-Based Retrieval n Audio-Score Alignment n Music Fingerprinting 2 Carnegie

Week 10A Query-by-Humming and Music Fingerprinting Roger B. Dannenberg Professor of Computer Science, Art and Music Carnegie Mellon University Overview n Melody-Based Retrieval n Audio-Score Alignment n Music Fingerprinting 2 Carnegie Mellon

658 views • 29 slides
Preference Policy Recommendations PREPARED BY Heidi Aggeler, Managing Director Denver, Colorado

Preference Policy Recommendations PREPARED BY Heidi Aggeler, Managing Director Denver, Colorado

City and County of Denver Preference Policy Recommendations PREPARED BY Heidi Aggeler, Managing Director Denver, Colorado 80220 970.880.1415 x102 heidi@rootpolicy.com Considerations for a Preference Policy Primary goal: To foster stable

585 views • 17 slides
Preference Litigation ASK LLP What is a preference? A preference is a payment made by an

Preference Litigation ASK LLP What is a preference? A preference is a payment made by an

Preference Litigation ASK LLP What is a preference? A preference is a payment made by an insolvent debtor that favors certain creditors over others. The causes of action to avoid and recover preferential payments are codified in 11

853 views • 39 slides
Planarity Embedding Embedding For a given graph G = ( V , E ) , an embedding (into R 2 ) assigns

Planarity Embedding Embedding For a given graph G = ( V , E ) , an embedding (into R 2 ) assigns

Planarity Embedding Embedding For a given graph G = ( V , E ) , an embedding (into R 2 ) assigns each vertex a coordinate and each edge a (not necessarily straight) line connecting the corresponding coordinates. 0 1 3 4 1 1 0 2 5 2 5

754 views • 18 slides
Graph Drawing Embedding Embedding For a given graph G = ( V , E ) , an embedding (into R 2 )

Graph Drawing Embedding Embedding For a given graph G = ( V , E ) , an embedding (into R 2 )

Graph Drawing Embedding Embedding For a given graph G = ( V , E ) , an embedding (into R 2 ) assigns each vertex a coordinate and each edge a (not necessarily straight) line connecting the corresponding coordinates. 0 1 3 4 1 1 0 2 5 2

541 views • 18 slides
Query Optimization Through the Looking Glass Some Lessons From Building an LLVM-Based Query

Query Optimization Through the Looking Glass Some Lessons From Building an LLVM-Based Query

Query Optimization Through the Looking Glass Some Lessons From Building an LLVM-Based Query Compiler Viktor Leis Technische Universitt Mnchen Introduction: Query Optimization HJ B cardinality cost INL B T SELECT ... estimation

679 views • 25 slides
Final Project: Music Preference Mackenzie McCreery, Karrie Chen, Alexander Solomon Introduction

Final Project: Music Preference Mackenzie McCreery, Karrie Chen, Alexander Solomon Introduction

Final Project: Music Preference Mackenzie McCreery, Karrie Chen, Alexander Solomon Introduction Physiological data Use has been increasing in User Experience (UX) research Its sensors record the involuntary physiological responses of

976 views • 30 slides
CAS CS 460/660 Introduction to Database Systems Query Evaluation II 1.1 Cost-based Query

CAS CS 460/660 Introduction to Database Systems Query Evaluation II 1.1 Cost-based Query

CAS CS 460/660 Introduction to Database Systems Query Evaluation II 1.1 Cost-based Query Sub-System Select * Queries From Blah B Where B.blah = blah Query Parser Query Optimizer Plan Plan Cost Catalog Manager Generator Estimator

711 views • 44 slides
Query Execu:on Declara:ve Query (SQL) We start from

Query Execu:on Declara:ve Query (SQL) We start from

SQL The Query Language R & G - Chapter 5 Based on Slides from UC Berkeley and book. 2 Query Execu:on Declara:ve Query (SQL) We start from

689 views • 34 slides
Query Execu:on Declara:ve Query (SQL) We start from

Query Execu:on Declara:ve Query (SQL) We start from

SQL III The Query Language R & G - Chapter 5 Based on Slides from UC Berkeley and book. Query Execu:on Declara:ve Query (SQL) We start from here

1.02k views • 64 slides
3. Preference Learning Techniques 4. Complexity of Preference Learning 5. Conclusions 1 ECAI

3. Preference Learning Techniques 4. Complexity of Preference Learning 5. Conclusions 1 ECAI

AGENDA 1. Preference Learning Tasks 2. Performance Assessment and Loss Functions 3. Preference Learning Techniques 4. Complexity of Preference Learning 5. Conclusions 1 ECAI 2012 Tutorial on Preference Learning | Part 5 | J. Frnkranz &

478 views • 6 slides
1 Logic-based Markup and Logic-based Markup and Query Language Query Language Eight

1 Logic-based Markup and Logic-based Markup and Query Language Query Language Eight

Overview Semantic Streams: a Framework for Composable Motivation Semantic Interpretation of Sensor What are Semantic Streams? Data Logic-based Markup and Query Language Kamin Whitehouse, Feng Zhao, and Jie Liu Query Processing

444 views • 5 slides
Personalized Recommendations for Music Genre Exploration Yu Liang, Martijn Willemsen HTI Group,

Personalized Recommendations for Music Genre Exploration Yu Liang, Martijn Willemsen HTI Group,

Personalized Recommendations for Music Genre Exploration Yu Liang, Martijn Willemsen HTI Group, Eindhoven University of Technology Jheronimus Academy of Data Science Liang, Yu, and Martijn C. Willemsen. "Personalized Recommendations for

513 views • 26 slides
MUSIC THERAPY MUSIC THERAPY What is music therapy? Music therapy is simply the process of using

MUSIC THERAPY MUSIC THERAPY What is music therapy? Music therapy is simply the process of using

MUSIC THERAPY MUSIC THERAPY What is music therapy? Music therapy is simply the process of using music to change a persons behaviour or provide memories to improve clarity. What is music therapy? Music can alter moods to manage

996 views • 30 slides
A Preference-Based Bandit Framework for Personalized Recommendation Maryam Tavakol and Ulf

A Preference-Based Bandit Framework for Personalized Recommendation Maryam Tavakol and Ulf

A Preference-Based Bandit Framework for Personalized Recommendation Maryam Tavakol and Ulf Brefeld Paderborn, Nov 8, 2016 Introduction Personalized Recommendation Preference Learning Multi-armed bandits 2 Recommendation 3 Recommendation

140 views • 13 slides
Cocoa Demo: GUI-based Stk app. Xcode Interface Builder StkX 5/5/06, Music 3SI,

Cocoa Demo: GUI-based Stk app. Xcode Interface Builder StkX 5/5/06, Music 3SI,

Last Week... Music 3SI: Introduction to IDE (briefly) Audio/Multimedia App. VST Plug-in Programming Assignment 1 hints Week #5 - 5/5/2006 CCRMA, Department of Music Stanford University 5/5/06, Music 3SI, CCRMA, Stanford 1 2

537 views • 7 slides
Optimizing Join Enumeration in Transformation-based Query

Optimizing Join Enumeration in Transformation-based Query

Optimizing Join Enumeration in Transformation-based Query Optimizers ANIL SHANBHAG, S. SUDARSHAN IIT BOMBAY anils@mit.edu, sudarsha@cse.iitb.ac.in VLDB 2014 Query Optimization: Quick

576 views • 18 slides
SAMPLING-BASED QUERY RE- OPTIMIZATION Wentao Wu Microsoft Research Background 2 Query

SAMPLING-BASED QUERY RE- OPTIMIZATION Wentao Wu Microsoft Research Background 2 Query

1 SAMPLING-BASED QUERY RE- OPTIMIZATION Wentao Wu Microsoft Research Background 2 Query optimization remains challenging despite of decades of efforts and progresses. Cardinality estimation is the key challenge. Selectivity of

400 views • 28 slides
Generating Computer Music from Skeletal Notation for Carnatic Music Compositions (M.

Generating Computer Music from Skeletal Notation for Carnatic Music Compositions (M.

2nd Comp-Music Workshop, Istanbul, 12-13 July, 2012 Generating Computer Music from Skeletal Notation for Carnatic Music Compositions (M. Subramanian) manianms@yahoo.com (Click here for a Web based presentation with Audio links) Genererating

331 views • 21 slides
A Generic Mapping-based Query Translation A Generic Mapping-based Query Translation from SPARQL

A Generic Mapping-based Query Translation A Generic Mapping-based Query Translation from SPARQL

A Generic Mapping-based Query Translation A Generic Mapping-based Query Translation from SPARQL to Various Target Database Query from SPARQL to Various Target Database Query Languages Languages F. Michel, C. Faron-Zucker, J. Montagnat I3S

668 views • 31 slides
Preference Defenses: New Value Ordinary Preference Defenses: New Value, Ordinary Course and

Preference Defenses: New Value Ordinary Preference Defenses: New Value, Ordinary Course and

Presenting a live 90 minute webinar with interactive Q&A Preference Defenses: New Value Ordinary Preference Defenses: New Value, Ordinary Course and Contemporaneous Exchange Managing the Audit Process, Mitigating Regulatory Sanctions, and

893 views • 52 slides
Ian Cross Centre for Music & Science Faculty of Music, Cambridge www.mus.cam.ac.uk/~cross

Ian Cross Centre for Music & Science Faculty of Music, Cambridge www.mus.cam.ac.uk/~cross

Ian Cross Centre for Music & Science Faculty of Music, Cambridge www.mus.cam.ac.uk/~cross Paper presented to the Philological Society QMUL 9 November 2012 For most of the C20th, much writing and thinking about music was based on the

611 views • 22 slides

More recommend