Introduction

Matrix and tensor factorizations have a long history in machine learning. Being able to recover latent factors in the data and flexible enough to accommodate a large set of constraints and regularizations, matrix and tensor factorization methods have found several applications in computer vision problems, providing a natural framework to handle the inherently complex structure of visual data (e.g., spatial and temporal dimensions in videos) and their multi-aspect, multimodal, and heterogeneous nature (e.g., RGB and depth images of the same object). Besides that, recent renaissance of tensor methods is furthermore attributed to current advances on the development of scalable algorithms for tensor operations and novel models through tensor representations that have deemed successful in unsupervised learning of latent variable models and dictionaries, uncovering high-order relations in the data, training deep neural networks, and explaining some of their theoretical aspects.

These progresses, along with industry solutions such as Google TensorFlow, Torch, and Tensor Processing Unit, trigger new directions and problems towards matrix and tensor methods in computer vision. The workshop aims to foster discussion, discovery, and dissemination of research activities and outcomes in this area and encourages breakthroughs. We will bring together researchers in theories and applications who are interested in analysis and factorization of tensors and matrices as second-order tensors and development of tensor-based algorithms for computer vision tasks. We will also invite researchers from related areas, such as numerical linear algebra, high-performance computing, deep learning, data analysis, among others, to contribute to this workshop. We believe that this workshop can foster new directions, closer collaborations and novel applications. We also expect a deeper conversation regarding why learning with tensors at current stage is important, where it is useful, what tensor computation software and hardware work well in practice and, how we can progress further with interesting research directions and open problems.

Program

08:40-08:50 Gathering and Welcome
Session 1 (8:50 - 10:40)
08:50-09:20 Invited Talk 1: Tensor decompositions for big multi-aspect data analytics Vagelis Papalexakis (University of California Riverside)
09:20-09:40 Multilevel Approximate Robust Principal Component Analysis Vahan Hovhannisyan, Yannis Panagakis, Panos Parpas, Stefanos Zafeiriou
09:40-10:00 Factorized Convolutional Neural Networks Min Wang, Baoyuan Liu, Hassan Foroosh
10:00-10:20 Factorization Approach for Enabling Structure-from-Motion/SLAM Using Integer Arithmetic Nilesh A Ahuja, Mahesh Subedar, Yeongseon Lee, Omesh Tickoo
10:20-10:40 Accurate Structure Recovery via Weighted Nuclear Norm: A Low Rank Approach to Shape-from-Focus Prashanth Kumar G, Rajiv Sahay
10:40-11:00 Break
Session 2 (11:00 - 12:40)
11:00-11:30 Invited Talk 2: Convolutional Dictionary Learning through Tensor Factorization Furong Huang (University of Maryland College Park)
11:30-12:00 Invited Talk 3: Shared Space Component Analysis Mihalis Nicolaou (Goldsmiths, University of London)
12:00-12:30 Tensor Learning in Python with TensorLy Jean Kossaifi (Imperial College London)
12:30-12:40 Concluding Remarks

Invited speakers

Yannis

Evangelos (Vagelis) Papalexakis is an Assistant Professor of the CSE Dept. at University of California Riverside. He received his PhD degree at the School of Computer Science at Carnegie Mellon University (CMU). Prior to CMU, he obtained his Diploma and MSc in Electronic & Computer Engineering at the Technical University of Crete, in Greece. Broadly, his research interests span the fields of Data Mining, Machine Learning, and Signal Processing. His research involves designing scalable algorithms for mining large multi-aspect datasets, with specific emphasis on tensor factorization models, and applying those algorithms to a variety of real world multi-aspect data problems. His work has appeared in KDD, ICDM, SDM, ECML-PKDD, WWW, PAKDD, ICDE, ICASSP, IEEE Transactions of Signal Processing, and ACM TKDD. He has a best student paper award at PAKDD'14, finalist best papers for SDM'14 and ASONAM'13 and he was a finalist for the Microsoft PhD Fellowship and the Facebook PhD Fellowship. Besides his academic experience, he has industrial research experience working at Microsoft Research Silicon Valley during the summers of 2013 and 2014 and Google Research during the summer of 2015. Finally, his doctoral dissertation received the 2017 SIGKDD Doctoral Dissertation Award (runner up).

Furong

Furong Huang is an assistant professor of computer science at University of Maryland College Park. Huang’s research focuses on machine learning, high-dimensional statistics and distributed algorithms—both the theoretical analysis and practical implementation of parallel spectral methods for latent variable graphical models. Some applications of her research include developing fast detection algorithms to discover hidden and overlapping user communities in social networks, learning convolutional sparse coding models for understanding semantic meanings of sentences and object recognition in images, healthcare analytics by learning a hierarchy on human diseases for guiding doctors to identify potential diseases afflicting patients, and more. Huang recently completed a postdoctoral position at Microsoft Research in New York.

Mihalis

Mihalis A. Nicolaou is a Lecturer (US equiv. Assistant Professor) at the Department of Computing at Goldsmiths, University of London and an Honorary Research Fellow with the Department of Computing at Imperial College London. He obtained his PhD from the same department at Imperial under an EPSRC Doctoral Training Award, while he completed his undergraduate studies at the University of Athens, Greece. Mihalis' research interests span the areas of machine learning, computer vision and affective computing. He has received several awards for his research, including a Best Paper Award at the IEEE Conference on Automatic Face and Gesture Recognition (AFGR). Mihalis has co-organized several workshops and special sessions in venues such as CVPR, while he served as a guest Associate Editor for the IEEE Transactions on Affective Computing. He has co-authored several papers that appear in top venues in the field such as IEEE T-PAMI, IEEE T-NNLS, CVPR, ECCV, ICCV and ECML.

Call for papers

We encourage discussions on recent advances in theory, algorithms, and applications of matrix and tensor factorization in computer vision. We are soliciting original contributions that address a wide range of theoretical and practical issues including, but not limited to:

 Advances in matrix and tensor factorization methods:

  • Matrix and tensor factorization methods for component analysis, dictionary learning, and latent variable models.
  • Matrix and tensor models with structural constraints (e.g., sparsity, low-rank, non-negativity etc.)
  • Matrix and tensor methods on non-Euclidean domains (e.g., kernels, manifolds, graphs etc.)
  • Robust to noise and outliers matrix and tensor factorizations
  • Mathematical optimization methods for matrix and tensor factorizations

Applications of matrix and tensor factorization methods to computer vision problems:

  • Factorization methods for rigid and non-rigid structure from motion, photometric stereo, and cameral calibration
  • Spatial and temporal segmentation and clustering of videos and image ensembles
  • Action and behavior analysis using tensors and tensor decompositions
  • Image enhancement, de-noising, and impainting using tensor methods
  • Tensor methods in medical imaging
  • Feature extraction using tensor methods
  • Tensor factorization for fusion of visual information with text, audio, and other modalities
  • Fast and scalable implementations of tensor methods for computer vision tasks
  • Software and hardware for tensors

Open and emerging research questions:

  • To role of invariance in learning with matrices and tensor. How to design matrix and tensor factorization that capture desirable invariances (e.g., invariant to geometric transformation, spatial and temporal deformations etc)?
  • Deep and non-linear matrix and tensor factorizations
  • Algorithms with theoretical guarantees for factorization methods

Important dates

Paper submission deadline: 1 August 2017
Author notification: 18 August 2017
Camera ready: 25 August 2017


Extended abstracts deadline: 15 September 2017
(2 to 4 pages without proceedings)


Workshop: 23 October 2017

Submission

Regular Papers

All submissions will be handled electronically via the conference’s CMT Website (https://cmt3.research.microsoft.com/TENSORCV2017).
The format for paper submission is the same as the ICCV main conference. The review will be double-blind. Each submission will be reviewed by at least two reviewers for originality, significance, clarity, soundness, relevance, and technical contents.

Papers that are not blind, or do not use the template, or have more than 10 pages (excluding references) will be rejected without review.

Extended Abstracts

Authors have the opportunity to submit extended abstracts (2 to 4 pages including references) that will be presented at the workshop. Extended abstracts will not be included in the proceedings in order to give authors the possibility to later submit their work to another venue.
The format of extended abstracts submission is the same as the ICCV main conference and will be handled electronically via the conference’s CMT Website ( https://cmt3.research.microsoft.com/TENSORCV2017 ).

The deadline for submission is September 29th (5:00 PM Pacific Time).

Organisers

Yannis

Yannis Panagakis is a Lecturer  (USA equivalent to Assistant Professor) in Computer Science at Middlesex University London and a Research Fellow at the Department of Computing, Imperial College London. He received his PhD and MSc degrees from the Department of Informatics, Aristotle University of Thessaloniki and his B.Sc. degree in Informatics and Telecommunication from the University of Athens, Greece. Yannis received various scholarships and awards for his studies and research, including the prestigious Marie-Curie Fellowship in 2013. He currently serves as an Associate Editor of Image and Vision Computing Journal. His current research interests include machine learning, signal processing, and mathematical optimization with applications to computer vision, human behavior analysis, and music information research. His work has been featured in top venues in the field, such as IEEE T-PAMI, TIP, IJCV as well as CVPR, and ICCV. Yannis will be the Workshops Chair in BMVC 2017.

Stefanos

Stefanos Zafeiriou is currently a Senior Lecturer (USA equivalent to Associate Professor) in Pattern Recognition/Statistical Machine Learning for Computer Vision with the Department of Computing, Imperial College London, London, U.K, and a Distinguishing Research Fellow with University of Oulu under Finish Distinguishing Professor Program. He was a recipient of the Prestigious Junior Research Fellowships from Imperial College London in 2011 to start his own independent research group. He was the recipient of the President's Medal for Excellence in Research Supervision for 2016. He has received various awards during his doctoral and post-doctoral studies. He currently serves as an Associate Editor of the IEEE Transactions on Cybernetics the Image and Vision Computing Journal. He has been a Guest Editor of over six journal special issues and co-organised over nine workshops/special sessions on face analysis topics in top venues, such as CVPR/FG/ICCV/ECCV (including two very successfully challenges run in ICCV’13 and ICCV’15 on facial landmark localisation/tracking). He has co-authored over 50 journal papers mainly on novel statistical machine learning methodologies applied to computer vision problems, such as 2-D/3- D face analysis, deformable object fitting and tracking, shape from shading, and human behaviour analysis, published in the most prestigious journals in his field of research, such as the IEEE T-PAMI, the International Journal of Computer Vision, the IEEE T-IP, the IEEE T-NNLS, the IEEE T-VCG, and the IEEE T-IFS, and many papers in top conferences, such as CVPR, ICCV, ECCV, ICML. His students are frequent recipients of very prestigious and highly competitive fellowships, such as the Google Fellowship, the Intel Fellowship, and the Qualcomm Fellowship. He is the General Chair of BMVC 2017.

Anima

Anima Anandkumar is currently a principal scientist at Amazon Web Services. She will be joining Caltech CMS department as a Bren endowed chair this summer.   Her research interests are in the areas of large-scale machine learning, non-convex optimization and high- dimensional statistics. In particular, she has been spearheading the development and analysis of tensor algorithms.  She is the recipient of several awards such as the Alfred. P. Sloan Fellowship, Microsoft Faculty Fellowship, Google research award, ARO and AFOSR Young Investigator Awards, NSF Career Award, Early Career Excellence in Research Award at UCI, Best Thesis Award from the ACM Sigmetrics society, IBM Fran Allen PhD fellowship, and several best paper awards. She has been featured in a number of forums such as the Quora ML session, Huffington post, Forbes, O’Reilly media, and so on.  She received her B.Tech in Electrical Engineering from IIT Madras in 2004 and her PhD from Cornell University in 2009. She was a postdoctoral researcher at MIT from 2009 to 2010,  an assistant professor at U.C. Irvine between 2010 and 2016, and a visiting researcher  at Microsoft Research New England in 2012 and 2014.

Jean

Jean Kossaifi is a Research Assistant and PhD student within the Department of Computing, Imperial College London, working as part of the iBUG group under Professor Maja Pantic's supervision. His current position follows the completion of an MSc in Advanced Computing, obtained with Distinction from Imperial College London, UK. In addition, Jean also holds a French Engineering Diploma/MSc in applied mathematics, finance and computing, obtained in parallel with a BSc in Advanced Mathematics.
His research interests are primarily focused on the areas of machine learning/tensor-learning, computer vision and pattern recognition, with applications in human-computer interaction, automatic non-verbal behaviour analysis and emotion recognition.