Baochen Sun

Microsoft Corporation

Redmond, WA

 

Contact

Email: bsun AT cs DOT uml DOT edu OR baochens AT gmail DOT com

 

 

 

 

I got my PhD and MS from University of Massachusetts Lowell (Advisor: Prof. Kate Saenko), and BEng from Shandong University, all in Computer Science. I am currently an applied scientist at Microsoft Corporation (Redmond, WA).

 

NEWS

Summer 2016: I defended my Ph.D. thesis on 08/05/2016.

12/2015: Our paper titled Return of Frustratingly Easy Domain Adaptation (Extended Abstract) has won the Best Paper Prize of the Transferring and Adapting Source Knowledge in Computer Vision (TASK-CV) workshop at ICCV'2015.

08/2015: I am co-organizing NIPS 2015 Workshop on Transfer and Multi-Task Learning: Trends and New Perspectives.

 

Research Interests

Computer Vision, Machine Learning, Domain Adaptation, Object Recognition, Deep Learning, Optimization

 

Publications

PhD thesis

Correlation Alignment for Domain Adaptation [Paper] [Slides]

 

Conference Papers

[1] B. Sun, J. Feng, and K. Saenko. Return of Frustratingly Easy Domain Adaptation. In AAAI, 2016 [Project page]

[2] X. Peng, B. Sun, K. Ali, and K. Saenko. Learning Deep Object Detectors from 3D Models. In ICCV, 2015 [Paper]

[3] B. Sun, K. Saenko. Subspace Distribution Alignment for Unsupervised Domain Adaptation. In BMVC, 2015 [Paper]

[4] B. Sun, K. Saenko. From Virtual to Reality: Fast Adaptation of Virtual Object Detectors to Real Domains. British Machine Vision Conference (BMVC), 2014 [Project page]

[5] B. Sun, X. Sun, Y. Wu, Y. Yin, and G. Yang. A New Algorithm for Generating Unique-Solution Sudoku. ICNC, 2008.

 

Journal Papers

[1] A. Chakrabarti, Y. Xiong, B. Sun, T. Darrell, D. Scharstein, T. Zickler, and K. Saenko. Modeling Radiometric Uncertainty for Vision with Tone-mapped Color Images. Pattern Analysis and Machine Intelligence, IEEE Transactions on, 2014 [Project page]

 

Workshop Papers

[1] B. Sun, J. Feng, and K. Saenko. Return of Frustratingly Easy Domain Adaptation. ICCV workshop on Transferring and Adapting Source Knowledge in Computer Vision (TASK-CV), 2015 [Poster] (Best Paper Prize)

[2] B. Sun, X. Peng, and K. Saenko. Generating Large Scale Image Datasets from 3D CAD Models. CVPR Workshop on the Future of Datasets in Vision, 2015 [Paper] [Poster] [Project page]

[3] X. Peng, B. Sun, K. Ali, and K. Saenko. Exploring Invariances in Deep Convolutional Neural Networks Using Synthetic Images. ICLR Workshop, 2015 [Paper] [Poster]

[4] B. Sun, K. Saenko. From Virtual to Reality: Fast Adaptation of Virtual Object Detectors to Real Domains. TASK-CV: Transferring and Adapting Source Knowledge in Computer Vision at ECCV 2014 [Poster]

[5] B. Schroeder, B. Sun, K. Saenko, and K. Ali. Deconstructing the Deformable Parts Model: Do More with Less. 9th Annual Workshop for Women in Machine Learning, 2014 [Paper] [Poster]

[6] B. Schroeder, B. Sun, K. Saenko, and K. Ali. Deconstructing the Deformable Parts Model: Do More with Less. Third International Workshop on Parts and Attributes at ECCV, 2014 [Paper] [Poster]

 

Invited Talks

[1] Correlation Alignment for Unsupervised Domain Adaptation. Google Research. Mountain View, CA, 2016.

[2] Correlation Alignment for Unsupervised Domain Adaptation. Google Research Cambridge. MA, 2016.

[3] Correlation Alignment for Unsupervised Domain Adaptation. Amazon. MA, 2016.

[4] Correlation Alignment for Unsupervised Domain Adaptation. Apple Inc. California, 2015.

[5] Correlation Alignment for Unsupervised Domain Adaptation. Vecna Technologies. Cambridge, MA, 2015.

[6] From Virtual to Reality: Fast Adaptation of Virtual Object Detectors to Real Domains. Amazon Fall Graduate Research Symposium. Seattle, 2014.

 

Honors and Awards

Best Paper Prize, Transferring and Adapting Source Knowledge in Computer Vision (TASK-CV) workshop at ICCV  2015

PM Raj Fund for Excellence in Computer Science  2014

Outstanding Student Scholarship (top 5-10%)  2005 - 2009

Scientific and Technological Innovation Scholarship (top 0.5%)  2008

1st Prize, National Post-Graduate Mathematical Contest in Modeling (top 4.1%)  2008

Dean's Scholarship (top 0.5%)  2007

5th Place, Robocup China Open 2007 (5/40)   2007

1st Prize, China Undergraduate Mathematical Contest in Modeling (top 2.1%)  2007

 

Research Experience

University of Massachusetts Lowell, Lowell, MA  01/2013 - 08/2016

Graduate Student Researcher with Prof. Kate Saenko

Researching on computer vision and machine learning

 

ICSI & EECS, UC Berkeley, Berkeley, CA  05/2015 - 08/2015

Visiting Student with Prof. Trevor Darrell and Stella Yu

Researching on domain adaptation, deep learning

 

University of Massachusetts Lowell, Lowell, MA  09/2010 - 12/2012

Graduate Student Researcher with Prof. Georges Grinstein

Researching on data visualization and data mining

 

Shandong University, Jinan, China  01/2008 - 2010/06

Undergraduate and Graduate Student Researcher with Prof. Xinshun Xu

Researching on machine learning and computational Intelligence

 

Teaching Experience

University of Massachusetts Lowell, Lowell, MA  09/2010 - 05/2013

Teaching Assistant for Machine Learning, Organization of Programming Languages, Topics in Bioinformatics, Visual Analytics, Operating Systems, and Computer Architecture.

 

Professional Activities

Organizer/Chair, NIPS 2015 Workshop on Transfer and Multi-Task Learning: Trends and New Perspectives

Reviewer/Program Committee, TPAMI, NIPS'16, ACCV'16, DeepVision Workshop'16

 

 

Projects

Correlation Alignment for Unsupervised Domain Adaptation

 

CORAL

Unlike human learning, machine learning often fails to handle changes between training (source) and test (target) input distributions. Such domain shifts, common in practical scenarios, severely damage the performance of conventional machine learning methods. Supervised domain adaptation methods have been proposed for the case when the target data have labels, including some that perform very well despite being "frustratingly easy" to implement. However, in practice, the target domain is often unlabeled, requiring unsupervised adaptation. We propose a simple, effective, and efficient method for unsupervised domain adaptation called CORrelation ALignment (CORAL). CORAL minimizes domain shift by aligning the second-order statistics of source and target distributions, without requiring any target labels. Even though it is extraordinarily simple--it can be implemented in four lines of Matlab code--CORAL performs remarkably well in extensive evaluations on standard benchmark datasets. [AAAI 2016 Paper (preprint)]


 

Building ImageNet in One Day

dataset

Datasets power computer vision research and drive breakthroughs. Larger and larger datasets are needed to better utilize the exponentially increasing computing power. However, datasets generation is both time consuming and expensive as human beings are required for image labelling. Human labelling cannot scale well. How can we generate larger image datasets easier and faster? In this paper, we provide a new approach for large scale datasets generation. We generate images from 3D object models directly. The large volume of freely available 3D CAD models and mature computer graphics techniques make generating large scale image datasets from 3D models very efficient. As little human effort involved in this process, it can scale very well. Rather than releasing a static dataset, we will also provide a software library for dataset generation so that the computer vision community can easily extend or modify the datasets accordingly. [Project Page]


 

Exploring Invariances In Deep Convolutional Neural Networks Using Synthetic Images

Description: Macintosh HD:Users:baochens:github:ICLR2015-VCNN:images:VCNN2.pngDeep convolutional neural networks learn extremely powerful image representations, yet most of that power is hidden in the millions of deep-layer parameters. What exactly do these parameters represent? Recent work has started to analyze CNN representations, finding that, e.g., they are invariant to some 2D transformations, but are confused by particular types of image noise. In this project, we delve deeper and ask: how invariant are CNNs to object-class variations caused by 3D shape, pose, and photorealism? These invariance properties are difficult to analyze using traditional data, so we propose an approach that renders synthetic data from freely available 3D CAD models. Using our approach we can easily generate an infinite amount of training images for almost any object. We explore the invariance of CNNs to various intra-class variations by simulating different rendering conditions, with surprising findings. Based on these results, we propose an optimal synthetic data generation strategy for training object detectors from CAD models. We show that our Virtual CNN approach significantly outperforms previous methods for learning object detectors from synthetic data on the benchmark PASCAL VOC2007 dataset. [ICCV 2015 Paper] [ICLR 2015 Workshop Paper]


Probabilistic Derendering of Camera Tone-Mapped Images

Description: Macintosh HD:Users:baochens:Desktop:derendering.pngTo produce images that are suitable for display, tone-mapping is widely used in digital cameras to map linear color measurements into narrow gamuts with limited dynamic range. This introduces non-linear distortion that must be undone, through a radiometric calibration process, before computer vision systems can analyze such photographs radiometrically. This project considers the inherent uncertainty of undoing the effects of tone-mapping. We observe that this uncertainty varies substantially across color space, making some pixels more reliable than others. We introduce a model for this uncertainty and a method for fitting it to a given camera or imaging pipeline. Once fit, the model provides for each pixel in a tone-mapped digital photograph a probability distribution over linear scene colors that could have induced it. We demonstrate how these distributions can be useful for visual inference by incorporating them into estimation algorithms for a representative set of vision tasks. [Project page]


 

From Virtual to Reality: Fast Adaptation of Virtual Object Detectors to Real Domains

Description: Macintosh HD:Users:baochens:Dropbox:bmvc2014:images:overview.pngThe most successful 2D object detection methods require a large number of images annotated with object bounding boxes to be collected for training. We present an alternative approach that trains on virtual data rendered from 3D models, avoiding the need for manual labeling. Growing demand for virtual reality applications is quickly bringing about an abundance of available 3D models for a large variety of object categories. While mainstream use of 3D models in vision has focused on predicting the 3D pose of objects, we investigate the use of such freely available 3D models for multicategory 2D object detection. To address the issue of dataset bias that arises from training on virtual data and testing on real images, we propose a simple and fast adaptation approach based on decorrelated features. We also compare two kinds of virtual data, one rendered with real-image textures and one without. Evaluation on a benchmark domain adaptation dataset demonstrates that our method performs comparably to existing methods trained on large-scale real image domains. [Paper] [Extended Abstract] [Poster] [Slides]

 


Deconstructing the Deformable Parts Model: Do More with Less

Description: Macintosh HD:Users:baochens:Desktop:Screen Shot 2015-01-22 at 10.18.48 AM.pngThe deformable parts model (DPM) is a successful detection model that continues to achieve state-of-the-art results in object detection. While the model's success is attributed to the deformable parts, the proposed system has many other design choices such as the use of multi-resolution HOG features, the sampling of parts from high energy areas, the use of a mixture model, etc. Recently, there have been attempts at analyzing the contribution of these design choices more closely. We delve further into this analysis, performing a study of the effects of restricting the deformation model, the use of single-resolution filters versus multi-resolution filters, and the application of energy 'dropout' while sampling parts. Our results indicate that a better performance can be achieved with simpler cost-free deformation models. [Paper] [Poster]

 


 

WEAVE: Web-based Analysis and Visualization Environment

Description: Macintosh HD:Users:baochens:Desktop:Screen Shot 2015-01-22 at 10.24.43 AM.pngWeave is a new web-based visualization platform designed to enable visualization of any available data by anyone for any purpose. Weave is an application development platform supporting multiple levels of users - novice to advanced - as well as the ability to integrate, analyze and visualize data at 'nested' levels of geography, and to disseminate the results in a web page.  [Project page]