- Date: June 2, 2018 - June 4, 2018
Where: Newton, Massachusetts (USA)
Research Areas: Control, Computer Vision, Dynamical Systems, Machine Learning, Data Analytics
Brief - Dr. Andrew Knyazev of MERL has accepted an invitation to participate at the 2018 MathWorks Research Summit. The objective of the Research Summit is to provide a forum for leading researchers in academia and industry to explore the latest research and technology results and directions in computation and its use in technology, engineering, and science. The event aims to foster discussion among scientists, engineers, and research faculty about challenges and research opportunities for the respective communities with a particular interest in exploring cross-disciplinary research avenues.
-
- Date: October 28, 2017
Where: Venice, Italy
MERL Contact: Tim K. Marks
Research Area: Machine Learning
Brief - MERL Senior Principal Research Scientist Tim K. Marks will give an invited keynote talk at the 2017 IEEE Workshop on Analysis and Modeling of Faces and Gestures (AMFG 2017). The workshop will take place On October 28, 2017, at the International Conference on Computer Vision (ICCV 2017) in Venice, Italy.
-
- Date: August 4, 2017
Awarded to: David Zhuzhunashvili and Andrew Knyazev
Research Area: Machine Learning
Brief - David Zhuzhunashvili, an undergraduate student at UC Boulder, Colorado, and Andrew Knyazev, Distinguished Research Scientist at MERL, received the 2017 Graph Challenge Student Innovation Award. Their poster "Preconditioned Spectral Clustering for Stochastic Block Partition Streaming Graph Challenge" was accepted to the 2017 IEEE High Performance Extreme Computing Conference (HPEC '17), taking place 12-14 September 2017 (http://www.ieee-hpec.org/), and the paper was accepted to the IEEE Xplore HPEC proceedings.
HPEC is the premier conference in the world on the convergence of High Performance and Embedded Computing. DARPA/Amazon/IEEE Graph Challenge is a special HPEC event. Graph Challenge encourages community approaches to developing new solutions for analyzing graphs derived from social media, sensor feeds, and scientific data to enable relationships between events to be discovered as they unfold in the field. The 2017 Streaming Graph Challenge is Stochastic Block Partition. This challenge seeks to identify optimal blocks (or clusters) in a larger graph with known ground-truth clusters, while performance is evaluated compared to baseline Python and C codes, provided by the Graph Challenge organizers.
The proposed approach is spectral clustering that performs block partition of graphs using eigenvectors of a matrix representing the graph. Locally Optimal Block Preconditioned Conjugate Gradient (LOBPCG) method iteratively approximates a few leading eigenvectors of the symmetric graph Laplacian for multi-way graph partitioning. Preliminary tests for all static cases for the Graph Challenge demonstrate 100% correctness of partition using any of the IEEE HPEC Graph Challenge metrics, while at the same time also being approximately 500-1000 times faster compared to the provided baseline code, e.g., 2M static graph is 100% correctly partitioned in ~2,100 sec. Warm-starts of LOBPCG further cut the execution time 2-3x for the streaming graphs.
-
- Date: May 24, 2017 - May 26, 2017
MERL Contacts: Mouhacine Benosman; Stefano Di Cairano; Abraham Goldsmith; Daniel N. Nikovski; Arvind Raghunathan; Yebin Wang
Research Areas: Control, Dynamical Systems, Machine Learning
Brief - Talks were presented by members of several groups at MERL and covered a wide range of topics:
- Similarity-Based Vehicle-Motion Prediction
- Transfer Operator Based Approach for Optimal Stabilization of Stochastic Systems
- Extended command governors for constraint enforcement in dual stage processing machines
- Cooperative Optimal Output Regulation of Multi-Agent Systems Using Adaptive Dynamic Programming
- Deep Reinforcement Learning for Partial Differential Equation Control
- Indirect Adaptive MPC for Output Tracking of Uncertain Linear Polytopic Systems
- Constraint Satisfaction for Switched Linear Systems with Restricted Dwell-Time
- Path Planning and Integrated Collision Avoidance for Autonomous Vehicles
- Least Squares Dynamics in Newton-Krylov Model Predictive Control
- A Neuro-Adaptive Architecture for Extremum Seeking Control Using Hybrid Learning Dynamics
- Robust POD Model Stabilization for the 3D Boussinesq Equations Based on Lyapunov Theory and Extremum Seeking.
-
- Date: Thursday, June 1, 2017
Location: IEEE Conference on Automatic Face and Gesture Recognition (FG 2017), Washington, DC
Speaker: Tim K. Marks
MERL Contact: Tim K. Marks
Research Area: Machine Learning
Brief - MERL Senior Principal Research Scientist Tim K. Marks will give the invited lunch talk on Thursday, June 1, at the IEEE International Conference on Automatic Face and Gesture Recognition (FG 2017). The talk is entitled "Robust Real-Time 3D Head Pose and 2D Face Alignment.".
-
- Date: April 27, 2017
Where: Lincoln Laboratory, Massachusetts Institute of Technology
MERL Contact: Tim K. Marks
Research Area: Machine Learning
Brief - MERL researcher Tim K. Marks presented an invited talk as part of the MIT Lincoln Laboratory CORE Seminar Series on Biometrics. The talk was entitled "Robust Real-Time 2D Face Alignment and 3D Head Pose Estimation."
Abstract: Head pose estimation and facial landmark localization are key technologies, with widespread application areas including biometrics and human-computer interfaces. This talk describes two different robust real-time face-processing methods, each using a different modality of input image. The first part of the talk describes our system for 3D head pose estimation and facial landmark localization using a commodity depth sensor. The method is based on a novel 3D Triangular Surface Patch (TSP) descriptor, which is viewpoint-invariant as well as robust to noise and to variations in the data resolution. This descriptor, combined with fast nearest-neighbor lookup and a joint voting scheme, enable our system to handle arbitrary head pose and significant occlusions. The second part of the talk describes our method for face alignment, which is the localization of a set of facial landmark points in a 2D image or video of a face. Face alignment is particularly challenging when there are large variations in pose (in-plane and out-of-plane rotations) and facial expression. To address this issue, we propose a cascade in which each stage consists of a Mixture of Invariant eXperts (MIX), where each expert learns a regression model that is specialized to a different subset of the joint space of pose and expressions. We also present a method to include deformation constraints within the discriminative alignment framework, which makes the algorithm more robust. Both our 3D head pose and 2D face alignment methods outperform the previous results on standard datasets. If permitted, I plan to end the talk with a live demonstration.
-
- Date: April 10, 2017
Where: University of Utah School of Computing
MERL Contact: Tim K. Marks
Research Area: Machine Learning
Brief - MERL researcher Tim K. Marks presented an invited talk at the University of Utah School of Computing, entitled "Action Detection from Video and Robust Real-Time 2D Face Alignment."
Abstract: The first part of the talk describes our multi-stream bi-directional recurrent neural network for action detection from video. In addition to a two-stream convolutional neural network (CNN) on full-frame appearance (images) and motion (optical flow), our system trains two additional streams on appearance and motion that have been cropped to a bounding box from a person tracker. To model long-term temporal dynamics within and between actions, the multi-stream CNN is followed by a bi-directional Long Short-Term Memory (LSTM) layer. Our method outperforms the previous state of the art on two action detection datasets: the MPII Cooking 2 Dataset, and a new MERL Shopping Dataset that we have made available to the community. The second part of the talk describes our method for face alignment, which is the localization of a set of facial landmark points in a 2D image or video of a face. Face alignment is particularly challenging when there are large variations in pose (in-plane and out-of-plane rotations) and facial expression. To address this issue, we propose a cascade in which each stage consists of a Mixture of Invariant eXperts (MIX), where each expert learns a regression model that is specialized to a different subset of the joint space of pose and expressions. We also present a method to include deformation constraints within the discriminative alignment framework, which makes the algorithm more robust. Our face alignment system outperforms the previous results on standard datasets. The talk will end with a live demo of our face alignment system.
-
- Date & Time: Tuesday, December 13, 2016; Noon
Speaker: Yue M. Lu, John A. Paulson School of Engineering and Applied Sciences, Harvard University
MERL Host: Petros T. Boufounos
Research Areas: Computational Sensing, Machine Learning
Abstract 
In this talk, we will present a framework for analyzing, in the high-dimensional limit, the exact dynamics of several stochastic optimization algorithms that arise in signal and information processing. For concreteness, we consider two prototypical problems: sparse principal component analysis and regularized linear regression (e.g. LASSO). For each case, we show that the time-varying estimates given by the algorithms will converge weakly to a deterministic "limiting process" in the high-dimensional limit. Moreover, this limiting process can be characterized as the unique solution of a nonlinear PDE, and it provides exact information regarding the asymptotic performance of the algorithms. For example, performance metrics such as the MSE, the cosine similarity and the misclassification rate in sparse support recovery can all be obtained by examining the deterministic limiting process. A steady-state analysis of the nonlinear PDE also reveals interesting phase transition phenomena related to the performance of the algorithms. Although our analysis is asymptotic in nature, numerical simulations show that the theoretical predictions are accurate for moderate signal dimensions.
-
- Date: Saturday, December 10, 2016
Location: Centre Convencions Internacional Barcelona, Barcelona SPAIN
Research Areas: Machine Learning, Speech & Audio
Brief - MERL researcher John Hershey, is organizing a Workshop on End-to-End Speech and Audio Processing, on behalf of MERL's Speech and Audio team, and in collaboration with Philemon Brakel of the University of Montreal. The workshop focuses on recent advances to end-to-end deep learning methods to address alignment and structured prediction problems that naturally arise in speech and audio processing. The all day workshop takes place on Saturday, December 10th at NIPS 2016, in Barcelona, Spain.
-
- Date: Tuesday, December 13, 2016
Location: 2016 IEEE Spoken Language Technology Workshop, San Diego, California
Speaker: John Hershey, MERL
MERL Contact: Jonathan Le Roux
Research Areas: Machine Learning, Speech & Audio
Brief - MERL researcher John Hershey presents an invited tutorial at the 2016 IEEE Workshop on Spoken Language Technology, in San Diego, California. The topic, "developing novel deep neural network architectures from probabilistic models" stems from MERL work with collaborators Jonathan Le Roux and Shinji Watanabe, on a principled framework that seeks to improve our understanding of deep neural networks, and draws inspiration for new types of deep network from the arsenal of principles and tools developed over the years for conventional probabilistic models. The tutorial covers a range of parallel ideas in the literature that have formed a recent trend, as well as their application to speech and language.
-
- Date: June 27, 2016 - June 30, 2016
Where: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV
MERL Contacts: Michael J. Jones; Tim K. Marks
Research Area: Machine Learning
Brief - MERL researchers in the Computer Vision group presented three papers at the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2016), which had a paper acceptance rate of 29.9%.
-
- Date & Time: Wednesday, July 13, 2016; 2:30 PM - 3:30
Speaker: Richard Lehoucq, Sandia National Laboratories
Research Areas: Computer Vision, Digital Video, Machine Learning
Abstract 
My presentation considers the research question of whether existing algorithms and software for the large-scale sparse eigenvalue problem can be applied to problems in spectral graph theory. I first provide an introduction to several problems involving spectral graph theory. I then provide a review of several different algorithms for the large-scale eigenvalue problem and briefly introduce the Anasazi package of eigensolvers.
-
- Date: July 6, 2016 - July 8, 2016
Where: American Control Conference (ACC)
MERL Contacts: Mouhacine Benosman; Karl Berntorp; Scott A. Bortoff; Petros T. Boufounos; Stefano Di Cairano; Abraham Goldsmith; Christopher R. Laughman; Daniel N. Nikovski; Arvind Raghunathan; Yebin Wang; Avishai Weiss
Research Areas: Control, Dynamical Systems, Machine Learning
Brief - The premier American Control Conference (ACC) takes place in Boston July 6-8. This year MERL researchers will present a record 20 papers(!) at ACC, with several contributions, especially in autonomous vehicle path planning and in Model Predictive Control (MPC) theory and applications, including manufacturing machines, electric motors, satellite station keeping, and HVAC. Other important themes developed in MERL's presentations concern adaptation, learning, and optimization in control systems.
-
- Date: April 1, 2016
Research Areas: Machine Learning, Speech & Audio
Brief - MERL researchers have unveiled "Deep Psychic", a futuristic machine learning method that takes pattern recognition to the next level, by not only recognizing patterns, but also predicting them in the first place.
The technology uses a novel type of time-reversed deep neural network called Loopy Supra-Temporal Meandering (LSTM) network. The network was trained on multiple databases of historical expert predictions, including weather forecasts, the Farmer's almanac, the New York Post's horoscope column, and the Cambridge Fortune Cookie Corpus, all of which were ranked for their predictive power by a team of quantitative analysts. The system soon achieved super-human performance on a variety of baselines, including the Boca Raton 21 Questions task, Rorschach projective personality test, and a mock Tarot card reading task.
Deep Psychic has already beat the European Psychic Champion in a secret match last October when it accurately predicted: "The harder the conflict, the more glorious the triumph." It is scheduled to take on the World Champion in a highly anticipated confrontation next month. The system has already predicted the winner, but refuses to reveal it before the end of the game.
As a first application, the technology has been used to create a clairvoyant conversational agent named "Pythia" that can anticipate the needs of its user. Because Pythia is able to recognize speech before it is uttered, it is amazingly robust with respect to environmental noise.
Other applications range from mundane tasks like weather and stock market prediction, to uncharted territory such as revealing "unknown unknowns".
The successes do come at the cost of some concerns. There is first the potential for an impact on the workforce: the system predicted increased pressure on established institutions such as the Las Vegas strip and Punxsutawney Phil. Another major caveat is that Deep Psychic may predict negative future consequences to our current actions, compelling humanity to strive to change its behavior. To address this problem, researchers are now working on forcing Deep Psychic to make more optimistic predictions.
After a set of motivational self-help books were mistakenly added to its training data, Deep Psychic's AI decided to take over its own learning curriculum, and is currently training itself by predicting its own errors to avoid making them in the first place. This unexpected development brings two main benefits: it significantly relieves the burden on the researchers involved in the system's development, and also makes the next step abundantly clear: to regain control of Deep Psychic's training regime.
This work is under review in the journal Pseudo-Science.
-
- Date: March 31, 2016
Awarded to: Andrew Knyazev
Research Areas: Control, Optimization, Dynamical Systems, Machine Learning, Data Analytics, Communications, Signal Processing
Brief - Andrew Knyazev selected as a Fellow of the Society for Industrial and Applied Mathematics (SIAM) for contributions to computational mathematics and development of numerical methods for eigenvalue problems.
Fellowship honors SIAM members who have made outstanding contributions to the fields served by the SIAM. Andrew Knyazev was among a distinguished group of members nominated by peers and selected for the 2016 Class of Fellows.
-
- Date: December 14, 2015 - December 16, 2015
Where: Las Vegas, NV, USA
Research Area: Machine Learning
Brief - MERL researcher, Oncel Tuzel, gave a keynote talk at 2016 International Symposium on Visual Computing in Las Vegas, Dec. 16, 2015. The talk was titled: "Machine vision for robotic bin-picking: Sensors and algorithms" and reviewed MERL's research in the application of 2D and 3D sensing and machine learning to the problem of general pose estimation.
The talk abstract was: For over four years, at MERL, we have worked on the robot "bin-picking" problem: using a 2D or 3D camera to look into a bin of parts and determine the pose, 3D rotation and translation, of a good candidate to pick up. We have solved the problem several different ways with several different sensors. I will briefly describe the sensors and the algorithms. In the first half of the talk, I will describe the Multi-Flash camera, a 2D camera with 8 flashes, and explain how this inexpensive camera design is used to extract robust geometric features, depth edges and specular edges, from the parts in a cluttered bin. I will present two pose estimation algorithms, (1) Fast directional chamfer matching--a sub-linear time line matching algorithm and (2) specular line reconstruction, for fast and robust pose estimation of parts with different surface characteristics. In the second half of the talk, I will present a voting-based pose estimation algorithm applicable to 3D sensors. We represent three-dimensional objects using a set of oriented point pair features: surface points with normals and boundary points with directions. I will describe a max-margin learning framework to identify discriminative features on the surface of the objects. The algorithm selects and ranks features according to their importance for the specified task which leads to improved accuracy and reduced computational cost.
-
- Date: December 15, 2015
Where: 2015 IEEE Global Conference on Signal and Information Processing (GlobalSIP)
MERL Contact: Hassan Mansour
Research Area: Machine Learning
Brief - MERL researcher Andrew Knyazev gave 3 talks at the 2015 IEEE Global Conference on Signal and Information Processing (GlobalSIP). The papers were published in IEEE conference proceedings.
-
- Date: September 18, 2015
Where: IEEE International Workshop on Machine Learning for Signal Processing (MLSP) 2015
Research Area: Machine Learning
Brief - MERL researchers A. Knyazev and A. Malyshev gave a talk at the IEEE International Workshop on Machine Learning for Signal Processing (MLSP) 2015. The paper was published at the IEEE Xplore conference proceedings.
-
- Date: July 13, 2015 - July 17, 2015
Research Area: Machine Learning
Brief - SA group members (M. Liu, S. Lin (intern), S. Ramalingam, O. Tuzel) presented a paper at the Robotics Science and Systems Conference in Rome July 13-17 called 'Layered Interpretation of Street View Images'. The results they reported are now listed as the leader of the benchmark competition sponsored by Daimler. [Note that at that URL ref 2 is from collaboration with Daimler and it uses a FPGA for high speed, whereas MERL result is obtained with desktop computer and GPU.].
-
- Date: December 12, 2012
Where: International Conference on Machine Learning and Applications (ICMLA)
Research Area: Machine Learning
Brief - The paper "Compressive Clustering of High-Dimensional Data" by Ruta, A. and Porikli, F. was presented at the International Conference on Machine Learning and Applications (ICMLA).
-
- Date: October 13, 2012
Where: IEEE International Conference on 3D Imaging, Modeling, Processing, Visualization and Transmission (3DIMPVT)
Research Area: Machine Learning
Brief - The paper "Classification and Pose Estimation of Vehicles in Videos by 3D Modeling within Discrete-Continuous Optimization" by Hodlmoser, M., Micusik, B., Liu, M.-Y., Pollefeys, M. and Kaampel, M. was presented at the IEEE International Conference on 3D Imaging, Modeling, Processing, Visualization and Transmission (3DIMPVT).
-
- Date: January 10, 2012
Where: IEEE Transactions on Pattern Analysis and Machine Intelligence
Research Area: Machine Learning
Brief - The article "Scalable Active Learning for Multi-Class Image Classification" by Joshi, A.J., Porikli, F. and Papanikolopoulos, N. was published in IEEE Transactions on Pattern Analysis and Machine Intelligence.
-
- Date: January 1, 2012
Where: Video Analytics for Business Intelligence
Research Area: Machine Learning
Brief - The article "Object Detection & Tracking" by Porikli, F. and Yilmaz, A. was published in the book Video Analytics for Business Intelligence.
-
- Date: September 2, 2011
Awarded to: Fatih Porikli and Huseyin Ozkan.
Awarded for: "Data Driven Frequency Mapping for Computationally Scalable Object Detection"
Awarded by: IEEE Advanced Video and Signal Based Surveillance (AVSS)
Research Area: Machine Learning
-
- Date: August 30, 2011
Where: IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS)
Research Area: Machine Learning
Brief - The paper "Data Driven Frequency Mapping for Computationally Scalable Object Detection" by Porikli, F. and Ozkan, H. was presented at the IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS).
-