Meta tags:

Headings (most frequently used words):

research, tom, minka, papers, software, teaching, and, tutorials, phd, computer, science, formerly, at, microsoft, cambridge, uk,

Text of the page (most frequently used words):
and (12), bayesian (10), inference (8), the (7), methods (6), mit (5), statistical (5), learning (5), for (5), this (5), software (4), cmu (4), data (4), computer (4), with (3), can (3), also (3), spring (3), research (3), have (3), models (3), vision (3), text (3), retrieval (3), page (2), see (2), january (2), 1997 (2), pattern (2), recognition (2), 2001 (2), approaches (2), 2003 (2), 2002 (2), mining (2), toolbox (2), matlab (2), papers (2), most (2), algorithms (2), into (2), first (2), determine (2), model (2), step (2), non (2), what (2), you (2), that (2), algorithm (2), make (2), machine (2), tom (2), minka (2), associated, paper, found, its, abstract, find, linkedin, curriculum, vitae, 1996, programming, language, exploration, patterns, reading, group, 1998, discovery, graphics, visualization, fall, glossary, teaching, tutorials, generalized, poissons, drawing, maps, bayes, point, machines, fitting, dirichlet, distributions, fastfit, lightspeed, tips, accelerating, worked, library, graphical, message, passing, went, infer, net, applied, matchmaking, online, multiplayer, games, trueskill, truematch, requires, good, taken, two, different, order, appropriate, developed, visualizing, taught, classes, second, approach, analyze, successful, assumptions, would, lead, those, reverse, engineering, process, usually, quite, instructive, improving, recovered, improve, their, results, visualize, makes, special, takes, account, all, possible, states, nature, not, just, one, likely, glance, seems, require, lot, computation, addressed, issue, developing, new, computational, including, obtain, benefits, typically, small, additional, cost, over, expectation, propagation, easier, understand, written, which, illustrate, important, problems, work, field, develop, efficient, use, goal, standard, tool, processing, information, formerly, microsoft, cambridge, phd, science,

Text of the page (random words):
tom minka tom minka phd computer science formerly at microsoft research cambridge uk hi i work in the field of bayesian statistical inference and i develop efficient algorithms for use in machine learning computer vision text retrieval and data mining my goal is to make bayesian inference a standard tool for processing information to make bayesian inference easier to understand i ve written papers which illustrate bayesian methods on important problems in machine learning computer vision and text retrieval what makes bayesian inference special is that it takes into account all possible states of nature not just the one that is the most likely at first glance this seems to require a lot of computation i ve addressed this issue by developing new computational methods including the expectation propagation algorithm with this algorithm you can obtain the benefits of bayesian inference with a typically small additional cost over non bayesian methods bayesian inference also requires good models i ve taken two different approaches to this the first is to visualize data in order to determine an appropriate model i have developed step by step methods for visualizing data taught in my classes at cmu my second approach is to analyze successful non bayesian methods in computer vision and text retrieval and determine what model assumptions would lead to those methods this reverse engineering process is usually quite instructive and by improving the recovered models you can improve on their results i have applied bayesian inference to matchmaking in online multiplayer games see truematch and trueskill i have also worked on infer net a software library for inference in graphical models most of my research in message passing algorithms went into it research papers software tips on accelerating matlab lightspeed matlab toolbox fastfit toolbox for fitting dirichlet distributions bayes point machines drawing maps generalized poissons software associated with a paper can be found on its abstract page teaching and tutorials a statistical learning pattern recognition glossary data mining at cmu fall 2003 2002 and 2001 statistical graphics and visualization at cmu spring 2003 and 2002 statistical approaches to learning and discovery at cmu spring 2001 pattern recognition at mit spring 1998 statistical learning reading group at mit 1997 software patterns at mit january 1997 programming language exploration at mit january 1996 also see my mit page my curriculum vitae find me on linkedin