Bit Twiddling Hack By Sean Eron Anderson seander@cs.stanford.edu Individually, the code snippets here are in the public domain (unless otherwise noted) — feel free to use them however you please. The aggregate collection and descriptions are © 1997-2005 Sean Eron Anderson. The code and descriptions are distributed in the hope that they will be useful, but WITHOUT ANY WARRANTY and without even the implied warranty of merchantability or fitness for a particular purpose. Contents About the operation counting methodology When totaling the number of operations for algorithms here, any C operator is counted as one operation. Compute the sign of an integer The last expression above evaluates to sign = v >> 31 for 32-bit integers. Alternatively, if you prefer the result be either -1 or +1, then use: sign = +1 | (v >> (sizeof(int) * CHAR_BIT - 1)); // if v < 0 then -1, else +1 On the other hand, if you prefer the result be either -1, 0, or +1, then use: sign = (v ! Detect if two integers have opposite signs f = v && !
Summary of course Machine Learning by Andrew Ng on Coursera – luckycallor This is my summary of course Machine Learning by Andrew Ng on Coursera. You can have a reference here after finishing the course. I'm glad to communicate with you and learn from each other. If you find any mistakes in the article, I would appreciate it if you pointed them out. a pdf edition 1. For m examples with n features, we can use a matrix X (with m rows and n columns) to describe the data, where row vector xi (with n+1 dimension including x0 ) represents an example, while column vector xj (with m dimension) represents a feature; or in matrix X, every element xij (row i, column j) represent the jth feature of ith example. For parameters, we use vector θ with n+1 elements to describe, where θj is correspond to xj . For labels, we use a vector y (with m elements) to represent, where element yi represent the label of ith example. And for every 1≤i≤m,xi0=0 . Hypothesis: Vector version: hθ=X∗θ Element version: hθ(xi)=n∑j=0xijθj Cost function: J(θ)=12mm∑i=1(hθ(xi)−yi)2 Gradient descent: θ=(XTX)−1XTy
CS345: Data Mining Data Mining Winter 2010 Course information: Instructors: Jure LeskovecOffice Hours: Wednesdays 9-10am, Gates 418 Anand RajaramanOffice Hours: Tuesday/Thursday 5:30-6:30pm (after the class in the same room) Room: Tuesday, Thursday 4:15PM - 5:30PM in 200-203 (History Corner). Teaching assistants: Abhishek Gupta (abhig@cs.stanford.edu). Roshan Sumbaly (rsumbaly@cs.stanford.edu). Staff mailing list: You can reach us at cs345a-win0910-staff@lists.stanford.edu Prerequisites: CS145 or equivalent. Materials: Readings have been derived from the book Mining of Massive Datasets. Students will use the Gradiance automated homework system for which a fee will be charged. You can see earlier versions of the notes and slides covering 2008/09 CS345a Data Mining. Requirements: There will be periodic homeworks (some on-line, using the Gradiance system), a final exam, and a project on web-mining. Projects: Course outline See Handouts for a list of topics and reading materials. Announcements: Important Dates
1.4. Support Vector Machines — scikit-learn 0.17 documentation The support vector machines in scikit-learn support both dense (numpy.ndarray and convertible to that by numpy.asarray) and sparse (any scipy.sparse) sample vectors as input. However, to use an SVM to make predictions for sparse data, it must have been fit on such data. For optimal performance, use C-ordered numpy.ndarray (dense) or scipy.sparse.csr_matrix (sparse) with dtype=float64. 1.4.1. SVC, NuSVC and LinearSVC are classes capable of performing multi-class classification on a dataset. SVC and NuSVC are similar methods, but accept slightly different sets of parameters and have different mathematical formulations (see section Mathematical formulation). As other classifiers, SVC, NuSVC and LinearSVC take as input two arrays: an array X of size [n_samples,n_features] holding the training samples, and an array y of class labels (strings or integers), size [n_samples]: After being fitted, the model can then be used to predict new values: >>> clf.predict([[2., 2.]])array([1]) 1.4.1.1. and . by
Research - Can Polling Location Influence STANFORD GRADUATE SCHOOL OF BUSINESS—What would you say influenced your voting decisions in the most recent local or national election? Political preferences? A candidate's stance on a particular issue? But Stanford Graduate School of Business researchers, doctoral graduates Jonah Berger and Marc Meredith, and S. It's hard to imagine that something as innocuous as polling location (e.g., school, church, or fire station) might actually influence voting behavior, but the Stanford researchers have discovered just that. Why might something like polling location influence voting behavior? Using data from Arizona's 2000 general election, Berger, Meredith, a visiting lecturer at MIT, and Wheeler discovered that people who voted in schools were more likely to support raising the state sales tax to fund education. This effect persisted even when the researchers controlled for—or removed the possibility of—other factors such as: Where voters lived.
CS 349: Data Mining, Search, and the World Tuesdays and Thursdays 4:15 - 5:30 in Bldg 370, Room 370 on the Main Quad Instructors: Sergey Brin and Lawrence Page Tues and Thurs 5:30 - 7:00 or by appointment. sergey@cs.stanford.edu and page@cs.stanford.edu Course Assistant: Diane Tang Gates 416: Mon - Wed 11:15 - 12:15 or by appointment. dtang@cs.stanford.edu Description Over the past two years there has been a close collaboration between the Data Mining Group (MIDAS) and the Digital Libraries Group at Stanford in the area of Web research. The topics of this class are data mining and information retrieval in the context of the World Wide Web. Prerequisites A strong knowledge of C. Very Tentative Syllabus Introduction: 1 Data Mining: 5 Publications of IBM's QUEST project 10/1 Market Basket (slides) R. Mailing List
Sergey Brin Ph.D. student in Computer Science at Stanford - sergey@cs.stanford.edu Research Currently I am at Google. In fall '98 I taught CS 349. Data Mining A major research interest is data mining and I run a meeting group here at Stanford. Extracting Patterns and Relations from the World Wide Web by Sergey Brin. World Wide Web Research on the Web seems to be fashionable these days and I guess I'm no exception. GNAT's This project involved indexing multidimensional data for near-neighbor searches. Near Neighbor Search in Large Metric Spaces by Sergey Brin. I worked on a project with Hector Garcia-Molina involving automated detection of copyright violations. Copy Detection Mechanisms for Digital Documents by Sergey Brin, James Davis, and Hector Garcia-Molina. Miscellaneous Photos My photo collection. HtmlTeX I have found existing tools to convert LaTeX into HTML a little frustrating so I wrote my own simple tool which relies mostly on a LaTeX style file. Pictures from Art Social Friends and Family End