Ines Chami

Ph.D. Candidate @Stanford University
Department: ICME
Email: chami@cs.stanford.edu

I recently graduated from my Ph.D. in ICME at Stanford University where I was advised by Prof. Christopher Ré. Prior to attending Stanford, I studied Mathematics and Computer Science at Ecole Centrale Paris.

My research interests include Representation Learning and non-Euclidean geometries (e.g., hyperbolic geometry). More specifically, I am interested in designing embedding models that can learn representations for complex relational structures such as graphs. I am particularly excited by applications in the field of Natural Language Processing, such as linking entities in Knowledge Graphs. During my studies, I had the chance to work on Question Answering at Microsoft AI and Research in 2017, and also spent the Summer of 2018 at Google Research, where I worked on graph-based representation learning.

During my free time, I enjoy surfing, practicing yoga and photography. I posted some of my pictures in the Photography section.

Keywords: Graph Representation Learning, Non-Euclidean Geometry, Knowledge Graphs

News

[06/21]: I graduated! My Ph.D. thesis Representation learning and algorithms in hyperbolic spaces is now available online.
[06/21]: Very grateful to be awarded the Gene Golub Dissertation Doctoral Dissertation Award.
[12/20]: I am giving a tutorial on hyperbolic embeddings and their uses in Machine Learning at the Neurips2020 DiffGeo4DL workshop. Check-out the video recording here!
[11/20]: I’ll be presenting our GraphEDM unified framework for graph representation learning at the 2020 Knowledge Connexions Conference!
[09/20]: I am very excited to be teaching the CS468 course "Non-Euclidean Methods in Machine Learning" @Stanford in the Fall 2020, together with Dr. Tolga Birdal and Prof. Leonidas Guibas. Our lecture slides are available here!
[09/20]: I am giving a talk for our paper on learning hyperbolic Knowledge Graph representations at the DOOCN 2020 Workshop.
[08/20]: I am co-organizing a Workshop at NeurIPS 2020 on Differential Geometry meets Deep Learning (DiffGeo4DL). Check out the Workshop webpage!
[04/20]: Our paper on low-dimensional Knowledge Graph embeddings is accepted at ACL 2020. PyTorch/Tensorflow implementations are now available!
[10/19]: Check out our new blog about Machine Learning in non-Euclidean spaces.
[09/19]: Excited to join Sujith Ravi's team at Google for a Fall internship!
[08/19]: Our paper on Hyperbolic Graph Convolutional Neural Networks is accepted at NeurIPS 2019!
[07/19]: I'm really honored to have received the Total Innovation Fellowship for the 2019/2020 academic year.
[05/19]: Our team achieved state-of-the-art performance on the MMTL GLUE benchmark! Check out the corresponding blog post here.

Publications

Hyperbolic Dimensionality Reduction via Horospherical Projections
International Conference on Machine Learning (ICML), 2021.
Ines Chami*, Albert Gu*, Dat Nguyen* and Christopher Ré.
[pdf] [code]

Tree Covers: An Alternative to Metric Embeddings
Differential Geometry meets Deep Learning Workshop @NeurIPS, 2020.
Roshni Sahoo, Ines Chami and Christopher Ré.
[pdf]

From Trees to Continuous Embeddings and Back: Hyperbolic Hierarchical Clustering
Advances in Neural Information Processing Systems (NeurIPS), 2020.
Ines Chami, Albert Gu, Vaggos Chatziafratis and Christopher Ré.
[pdf] [code] [video]

Machine Learning on Graphs: A Model and Comprehensive Taxonomy
Arxiv preprint, 2020.
Ines Chami, Sami Abu-El-Haija, Bryan Perozzi, Christopher Ré and Kevin Murphy.
[pdf] [Tensorflow code]

Low-Dimensional Hyperbolic Knowledge Graph Embeddings
Annual Conference of the Association for Computational Linguistics (ACL), 2020. [pdf]
Graph Representation Learning Workshop @NeurIPS, 2019. [pdf]
Ines Chami, Adva Wolf, Da-Cheng Juan, Frederic Sala, Sujith Ravi and Christopher Ré.
[Tensorflow code] [PyTorch code] [video]

Hyperbolic Graph Convolutional Neural Networks
Advances in Neural Information Processing Systems (NeurIPS), 2019.
Ines Chami*, Rex Ying*, Christopher Ré and Jure Leskovec.
[pdf] [code] [website]

Referring Relationships
IEEE International Conference on Computer Vision and Pattern Recognition (CVPR), 2018.
Ranjay Krishna*, Ines Chami*, Michael Bernstein and Li Fei-Fei.
[pdf] [code] [website] [video]

Abstract Meta Concept Features for Text-Illustration
ACM International Conference on Multimedia Retrieval (ICMR), 2017. (Oral Presentation)
Ines Chami*, Youssef Tamaaazousti* and Hervé Le Borne.
[pdf] [slides] [poster]

Image Annotation and Two Paths to Text-Illustration
CLEF (Working Notes), 2016.
Hervé Le Borne, Etienne Gadeski, Ines Chami, Thi Quynh Nhi Tran, Youssef Tamaaazousti, Alexandru Lucian Ginsca and Adrian Popescu.
[pdf]