Workshop

Machine Learning for Media Discovery

Erik Schmidt, Oriol Nieto, Fabien Gouyon, Yves Raimond, Katherine Kinnaird, Gert Lanckriet

Keywords:  Bandits    Recommender Systems    Content Discovery    Media    Multimedia    Music    Reinforcement Learning    Signal Processing    Machine Creativity  

Abstract:

The ever-increasing size and accessibility of vast media libraries has created a demand more than ever for AI-based systems that are capable of organizing, recommending, and understanding such complex data.

While this topic has received only limited attention within the core machine learning community, it has been an area of intense focus within the applied communities such as the Recommender Systems (RecSys), Music Information Retrieval (MIR), and Computer Vision communities. At the same time, these domains have surfaced nebulous problem spaces and rich datasets that are of tremendous potential value to machine learning and the AI communities at large.

This year's Machine Learning for Media Discovery (ML4MD) aims to build upon the five previous Machine Learning for Music Discovery editions at ICML, broadening the topic area from music discovery to media discovery. The added topic diversity is aimed towards having a broader conversation with the machine learning community and to offer cross-pollination across the various media domains.

One of the largest areas of focus in the media discovery space is on the side of content understanding. The recommender systems community has made great advances in terms of collaborative feedback recommenders, but these approaches suffer strongly from the cold-start problem. As such, recommendation techniques often fall back on content-based machine learning systems, but defining the similarity of media items is extremely challenging as myriad features all play some role (e.g., cultural, emotional, or content features, etc.). While significant progress has been made, these problems remain far from solved.

In addition, these complex data present many challenges beyond the development of machine learning systems to model and understand them. One of the largest challenges is scale. One example is commercial music libraries, which span into the tens of millions. However, user-generated content platforms such as YouTube and Pinterest have libraries stretching into the billions--a scale at which many of the traditional approaches discussed in the literature simply cannot perform.

On the other side of this problem sits the recent explosion of work in the area of Creative AI. Relevant examples include Google Magenta, Amazon's DeepComposer, who seek to develop algorithms capable of composing and performing completely original (and compelling) works of music. The same also happens in the world of visual media creation (e.g., DeepDream, Deep Fakes). Certain work in this area adds an interesting dimension to the conversation as understanding how content is created is a prerequisite to generating.

This workshop proposal is timely in that it will bridge these separate pockets of otherwise very related research. In addition to making progress on the challenges above, we hope to engage the wide AI and machine learning community with our rich problem space, and connect them with the many available datasets the community has to offer.

Chat is not available.

Timezone: »

Schedule

Sat 9:00 a.m. - 9:10 a.m.
Welcome Remarks (Welcome)
Sat 9:10 a.m. - 9:40 a.m.
Graph Neural Networks for Reasoning over Multimodal Content (Invited Talk)   
Jure Leskovec
Sat 9:40 a.m. - 10:00 a.m.
Novel Audio Embeddings for Personalized Recommendations on Newly Released Tracks (Accepted Talk)   
Beici Liang
Sat 10:00 a.m. - 10:20 a.m.
Musical Word Embedding: Bridging the Gap between Listening Contexts and Music (Accepted Talk)   
Seungheon Doh
Sat 10:20 a.m. - 11:00 a.m.

Both poster sessions will feature the full poster program.

Bach or Mock? A Grading Function for Chorales in the Style of J.S. Bach
Alexander Fang, Alisa Liu, Prem Seetharaman and Bryan Pardo
Topic: ML4MD Poster: Fang et al.
https://netflix.zoom.us/j/99893045607?pwd=RHRzeUZueGhYMVo1Rzk0V3NlSHV1UT09
Meeting ID: 998 9304 5607
Password: ML4MD

Incorporating Music Knowledge in Continual Dataset Augmentation for Music Generation
Alisa Liu, Alexander Fang, Gaëtan Hadjeres, Prem Seetharaman and Bryan Pardo
Topic: ML4MD Poster: Liu et al.
https://netflix.zoom.us/j/96351571989?pwd=UmtrUlpSSTk1UEhHUVlRV1NzL0xPQT09
Meeting ID: 963 5157 1989
Password: ML4MD

Artist biases in collaborative filtering for music recommendation
Andres Ferraro, Jae Ho Jeon, Biho Kim, Xavier Serra and Dmitry Bogdanov
Topic: ML4MD Poster: Ferraro et al.
https://netflix.zoom.us/j/91972151215?pwd=QllVZ2QxTUxmaCtWZmdyRks4UW55UT09
Meeting ID: 919 7215 1215
Password: ML4MD

Discovering X Degrees of Keyword Separation in a Fine Arts Collection
Arthur Flexer
Topic: ML4MD Poster: Flexer
https://netflix.zoom.us/j/94547050977?pwd=d3Z6NVJ6ZFRkV0lIWjVWcHdjd1lDUT09
Meeting ID: 945 4705 0977
Password: ML4MD

Generative Modelling for Controllable Audio Synthesis of Expressive Piano Performance
Hao Hao Tan, Yin-Jyun Luo and Dorien Herremans
Topic: ML4MD Poster: Tan et al.
https://netflix.zoom.us/j/94487939901?pwd=TXBFNU1Zd0FCZy9sQllzZktMUTU5QT09
Meeting ID: 944 8793 9901
Password: ML4MD

Web Interface for Exploration of Latent and Tag Spaces in Music Auto-Tagging
Philip Tovstogan, Xavier Serra and Dmitry Bogdanov
Topic: ML4MD Poster: Tovstogan et al.
https://netflix.zoom.us/j/91482530587?pwd=dmZnV1d5a2Z1RjZWa0VvSEF2SVBWUT09
Meeting ID: 914 8253 0587
Password: ML4MD

Cosine Similarity of Multimodal Content Vectors for TV Programmes
Saba Nazir, Taner Cagali, Chris Newell and Mehrnoosh Sadrzadeh
Topic: ML4MD Poster: Nazir et al.
https://netflix.zoom.us/j/99906337836?pwd=WnBBVFVxMEI5KzhhSUpEeFRRb0IyQT09
Meeting ID: 999 0633 7836
Password: ML4MD

Self-Correcting Non-Chronological Autoregressive Music Generation
Wayne Chi, Prachi Kumar, Suri Yaddanapudi, Rahul Suresh and Umut Isik
Topic: ML4MD Poster: Chi et al.
https://netflix.zoom.us/j/91455879996?pwd=VTUvK0szb093YnlGRTFCYWdaZEZKdz09
Meeting ID: 914 5587 9996
Password: ML4MD

Sat 11:00 a.m. - 11:30 a.m.
Graphs for music analysis (Invited Talk)   
Delia Fano Yela
Sat 11:30 a.m. - 11:50 a.m.
Deep Active Learning Toward Crisis-related Tweets Classification (Accepted Talk)   
Shiva Ebrahimi
Sat 11:50 a.m. - 12:20 p.m.
The Unsung Heroes of Music Recommendation: an Essay (Invited Talk)   
Matthias Mauch
Sat 12:20 p.m. - 1:00 p.m.
Lunch (Break)
Sat 1:00 p.m. - 1:30 p.m.
Beyond Being Accurate: Solving Real-World Recommendation Problems with Neural Modeling (Invited Talk)   
Ed Chi
Sat 1:30 p.m. - 1:50 p.m.
Character-focused Video Thumbnail Retrieval (Accepted Talk)   
Shervin Ardeshir
Sat 1:50 p.m. - 2:10 p.m.
HitPredict: Using Spotify Data to Predict Billboard Hits (Accepted Talk)   
Elena Georgieva
Sat 2:10 p.m. - 2:50 p.m.

Both poster sessions will feature the full poster program.

Bach or Mock? A Grading Function for Chorales in the Style of J.S. Bach
Alexander Fang, Alisa Liu, Prem Seetharaman and Bryan Pardo
Topic: ML4MD Poster: Fang et al.
https://netflix.zoom.us/j/99893045607?pwd=RHRzeUZueGhYMVo1Rzk0V3NlSHV1UT09
Meeting ID: 998 9304 5607
Password: ML4MD

Incorporating Music Knowledge in Continual Dataset Augmentation for Music Generation
Alisa Liu, Alexander Fang, Gaëtan Hadjeres, Prem Seetharaman and Bryan Pardo
Topic: ML4MD Poster: Liu et al.
https://netflix.zoom.us/j/96351571989?pwd=UmtrUlpSSTk1UEhHUVlRV1NzL0xPQT09
Meeting ID: 963 5157 1989
Password: ML4MD

Artist biases in collaborative filtering for music recommendation
Andres Ferraro, Jae Ho Jeon, Biho Kim, Xavier Serra and Dmitry Bogdanov
Topic: ML4MD Poster: Ferraro et al.
https://netflix.zoom.us/j/91972151215?pwd=QllVZ2QxTUxmaCtWZmdyRks4UW55UT09
Meeting ID: 919 7215 1215
Password: ML4MD

Discovering X Degrees of Keyword Separation in a Fine Arts Collection
Arthur Flexer
Topic: ML4MD Poster: Flexer
https://netflix.zoom.us/j/94547050977?pwd=d3Z6NVJ6ZFRkV0lIWjVWcHdjd1lDUT09
Meeting ID: 945 4705 0977
Password: ML4MD

Generative Modelling for Controllable Audio Synthesis of Expressive Piano Performance
Hao Hao Tan, Yin-Jyun Luo and Dorien Herremans
Topic: ML4MD Poster: Tan et al.
https://netflix.zoom.us/j/94487939901?pwd=TXBFNU1Zd0FCZy9sQllzZktMUTU5QT09
Meeting ID: 944 8793 9901
Password: ML4MD

Web Interface for Exploration of Latent and Tag Spaces in Music Auto-Tagging
Philip Tovstogan, Xavier Serra and Dmitry Bogdanov
Topic: ML4MD Poster: Tovstogan et al.
https://netflix.zoom.us/j/91482530587?pwd=dmZnV1d5a2Z1RjZWa0VvSEF2SVBWUT09
Meeting ID: 914 8253 0587
Password: ML4MD

Cosine Similarity of Multimodal Content Vectors for TV Programmes
Saba Nazir, Taner Cagali, Chris Newell and Mehrnoosh Sadrzadeh
Topic: ML4MD Poster: Nazir et al.
https://netflix.zoom.us/j/99906337836?pwd=WnBBVFVxMEI5KzhhSUpEeFRRb0IyQT09
Meeting ID: 999 0633 7836
Password: ML4MD

Self-Correcting Non-Chronological Autoregressive Music Generation
Wayne Chi, Prachi Kumar, Suri Yaddanapudi, Rahul Suresh and Umut Isik
Topic: ML4MD Poster: Chi et al.
https://netflix.zoom.us/j/91455879996?pwd=VTUvK0szb093YnlGRTFCYWdaZEZKdz09
Meeting ID: 914 5587 9996
Password: ML4MD

Sat 2:50 p.m. - 3:20 p.m.
Hit Song Prediction (Invited Talk)   
Eva Zangerle
Sat 3:20 p.m. - 3:40 p.m.
I know why you like this movie: Interpretable Efficient Mulitmodal Recommender (Accepted Talk)   
Barbara Rychalska
Sat 3:40 p.m. - 4:00 p.m.
Content-based Music Similarity with Siamese Networks (Accepted Talk)   
Joe O Cleveland