Computer Science Driver

MATTHEW D. FISHER

704-***-**** ** Olmsted Rd. Apt. 104, Stanford, CA 94305

********@********.*** http://graphics.stanford.edu/~mdfisher/projects.html

Education

****-******* **.*. ** Computer Science under Patrick Hanrahan, Stanford University, GPA 4.15/4.3

B.S. in Computer Science, California Institute of Technology, GPA 3.9/4.3

2003-2007

North Carolina School of Science and Mathematics

2001-2003

Skills

C/C++, Java, DirectX, OpenGL, HLSL, GLSL, WinDbg, Assembly, C#, Mathematica, MATLAB, Python

Work Experience

Stanford Stanford, CA

Context-Aware Scene Modeling September 2009 Present

Gathered very large scene databases from virtual worlds and online 3D model collections and analyzed this dataset

to learn the contextual relationship between models

This contextual information is used to assist both professional and casual artists as they model scenes by ordering

the set of models based on how likely they are to appear given the surrounding environment

Currently exploring other uses for this information, such as automatically decorating partially modeled scenes with

relevant objects, and suggesting replacement models for an object that has already been placed

Published a SIGGRAPH Asia 2010 and a SIGGRAPH 2011 paper based on this work

Stanford Stanford, CA

Japanese Text Translation for Tourists Winter 2009

Designed a program that uses machine learning to recognize Japanese or Chinese text in images and translate it into

English, targeting mobile devices and cameras

Allows users to correct character separation and transcription errors, focusing on users unfamiliar with Unicode

input in these languages or other aspects that can complicate translation ( vs., vs., vs., etc.)

Achieves better than 95% transcription accuracy on high-resolution input images (40x40 or larger pixels/character)

Considers the full probability distribution over each character when translating to robustly handle misclassified

characters

Stanford Stanford, CA

Camera-Only Robot Localization Spring 2009

Deployed a robot equipped with a three axis camera and both horizontal and vertical range scanners through a

crowded laboratory environment

Used the simultaneous localization and mapping (SLAM) algorithm on the horizontal scanner data to reconstruct a

2D map of the traversed environment and incorporated the information from the vertical scanner and cameras to

create a fully textured 3D map of the building

Designed an algorithm to enable robots that do not have range scanners and are equipped with only a camera to

determine their position and orientation in this environment using the generated 3D environment map

Stanford Stanford, CA

Micropolygon Rendering August 2008 January 2010

Modified the modern real-time graphics pipeline so that it scales better in the case of production quality scenes,

which are often characterized by polygons which are less than a pixel in size

Proposed and implemented a parametric patch tessellation algorithm that does not produce cracks, adapts well to

changing surface complexity, and permits a simple parallel implementation

Published the algorithm and presented it as first author at the SIGGRAPH Asia 2009 conference in Yokohama, Japan

Stanford Stanford, CA

Mesh Reconstruction from Images Fall 2007 Spring 2008

Implemented an algorithm to reconstruct a 3D mesh from a set of 2D images of an object whose key feature is to

iteratively project a candidate mesh back into the input images and converge to a stable and accurate result

GPU implementation reduced computation time from 30 minutes per image to less than 5 seconds per image, and

produces a fully textured and closed mesh

Achieved over 99% accuracy using 47 images on the standard Temple and Dino datasets

Caltech Pasadena, CA

Discrete Differential Geometry June 2006 June 2007

Collaborated with researchers from Microsoft Research and Caltech to develop a system for user-controlled design

of tangent vector fields on a surface based on discrete differential geometry

Uses texture synthesis as a compelling target application

First author on the associated paper which was presented at SIGGRAPH 2007

Microsoft Seattle, WA

DirectX SoftGPU Team Summer 2007

Developed a kernel driver that implements the Windows graphics driver model entirely in software and mounts

either on a VGA device or as a secondary monitor

Integrated this driver with the Microsoft WARP software renderer to enable optimized 2D and 3D graphics

Used this driver to test Windows driver model features, such as virtualized graphics card memory, that were under

development and did not yet have existing graphics cards or drivers that support these features

Microsoft Seattle, WA

DirectX Kernel Team Summers 2005, 2006

Developed GPUView, a comprehensive tool for investigating graphics performance issues that is still in use by

Microsoft and its hardware partners for the Windows Vista & Windows 7 Display Driver Model and ships with the

Windows 7 SDK

Wrote a patent called Kernel Event Visualization based around this work

Investigated and resolved several performance bottlenecks in the OS

Implemented a new kernel entry point for graphics drivers

Teaching Experience

CS148: Introduction to Computer Graphics and Imaging, Instructor Stanford University, Summer 2009

CS148: Introduction to Computer Graphics and Imaging, TA, Ranjitha Kumar Stanford University, Summer 2008

CS148: Introduction to Computer Graphics and Imaging, TA, Patrick Hanrahan Stanford University, Winter 2008

CS/CNS/EE 156b: Projects in Learning Systems, Lead TA, Yaser S. Abu-Mostafa Caltech, Winter 2006

CS/CNS/EE 156a: Learning Systems, Lead TA, Yaser S. Abu-Mostafa Caltech, Fall 2005

Publications

FISHER, M., SAVVA, M., AND HANRAHAN, P. Characterizing Structural Relationships in Scenes Using Graph Kernels. SIGGRAPH 2011.

FISHER, M., AND HANRAHAN, P. Context-Based Search for 3D Models. SIGGRAPH Asia 2010.

FISHER, M., FATAHALIAN, K., BOULOS, S., AKELEY, K., MARK, W., AND HANRAHAN, P. DiagSplit: Parallel, Crack-Free, Adaptive Tessellation

for Micropolygon Rendering. SIGGRAPH Asia 2009.

FISHER, M., SCHR DER, P., DESBRUN, M., AND HOPPE, H. 2007. Design of Tangent Vector Fields. SIGGRAPH 2007.

FISHER, M., SPRINGBORN, B., BOBENKO, A. I., AND SCHR DER, P. An Algorithm for the Construction of Intrinsic Delaunay Triangulations

with Applications to Digital Geometry Processing. Discrete Differential Geometry Course Notes, SIGGRAPH 2006.

Scholarships and Honors

Hertz Foundation Applied Science Fellowship Hertz Foundation, Eric Wepsic

Housner Award for Undergraduate Achievement California Institute of Technology

Upper Class Merit Award California Institute of Technology

Barry M. Goldwater Scholarship US Government

Bhansali prize for best undergraduate researcher in CS California Institute of Technology, CS Faculty

Fred V. and Marvis B. Maloney Scholarship California Institute of Technology

Richard Brewer prize for best physics research hurdle solution California Institute of Technology, Physics Faculty

Contact this candidate