background preloader

Machine Learning

Facebook Twitter

JAMA: Java Matrix Package. Background.

JAMA: Java Matrix Package

Michele Filannino. My name is Michele Filannino (yes, Michele with just one "L") and I am a PhD student at Centre for Doctoral Training in Computer Science in the University of Manchester.

Michele Filannino

The Manchester Centre for Doctoral Training in Computer Science proposes a new model of PhD training which combines the deep technical research study associated with the UK PhD with training and practical experience in creativity and innovation and scientific evaluation, and gives students experiences practicing their research skills by working with users from outside academia. It is the first and currently only EPSRC funded CDT in core computer science, and is led by Professor Steve Furber and Dr. Jonathan Shapiro. Download ManTIME DEMO is online! Stuff Research group. Programming-by-demonstration.org. Q+A - machine learning, natural language processing, artificial intelligence, text analysis, information retrieval, search, data mining, statistical modeling, and data visualization. MML Lab. (ML 1.1) Machine learning - overview and applications. Machine Learning. Christian Borgelt's Web Pages. BCView - Bayes Classifier Visualization Download Note: The table package contains some auxiliary programs for preprocessing data files.

Christian Borgelt's Web Pages

Description BCView is a program to visualize the numeric part (i.e., the part with numeric descriptive attributes) of a full or a naive Bayes classifier as they can be induced by the full and naive Bayes classifier programs of the Bayes package, of a cluster set as it can be induced by the fuzzy and probabilistic fuzzy clustering programs of the Cluster package, and of the hidden layer of a radial basis function network as it can be trained with the programs of the radial basis function network package. If you have trouble executing the program on Microsoft Windows, check whether you have the Microsoft Visual C++ Redistributable Packages for Visual Studio 2013 installed, as the C program was compiled with Microsoft Visual Studio 2013. Прогнозирование - Лекция 2. Корреляционный анализ и простая линейна... Free Science & Engineering software downloads.

Netflix algorithm: Prize Tribute Recommendation Algorithm in Python. Public:gaussian_mixture_models_em_algorithm_-_demo [juergen's work wiki] Here I generated some samples on the letters of the word “hello” and tried to learn a Gaussian Mixture Model (GMM) with n=1,2,…,100 components.

public:gaussian_mixture_models_em_algorithm_-_demo [juergen's work wiki]

I.e. describing the sample distribution with 1,2,…,100 Gaussians where each Gaussian is described by a mean 2D position a (full) 2×2 covariance matrix and a weight (remember: GMM := sum of weighted Gaussians) Conclusions: it roughly works the more gaussian you use in your GMM model, the longer the EM training takes but it is still “fast”, i.e. in the order of less or just a few seconds for all n (1,2,…,100) note! If the weight of a gaussian is 0.0 after learning: you get an invalid cluster center with location cx=-1.

BTW: the weight of each Gaussians is visualized by the “white-ness” of the corresponding ellipse's line or fill-color, respectively. Numenta, Inc. Grok is a cloud-based service that takes your data streams and generates actionable predictions in real time. Grok does this by finding complex patterns in your data and then using these patterns to make predictions about what will come next. You stream data into Grok; it returns a stream of predictions that can drive decisions and actions. Automated analytics lets you scale deployment Grok automates the time-consuming process of data analysis and predictive modeling. This empowers IT and business managers to deploy real-time prediction systems without access to analytics experts becoming a constraint.

Grok’s automation also allows you to model data and take action with greater granularity than conventional data analytics tools. Adapts in real time to changes in high-velocity data Grok excels at the analysis of streaming data. Enables automated action to maximize efficiency. Michael F Cohen. Papers Real-time Drawing Assistance through Crowdsourcing Alex Limpaecher, Nicolas Feltman, Adrien Treuille, and Michael F.

Michael F Cohen

Cohen SIGGRAPH 2013 UIST 2012 Best Paper Award Winner ! Cliplets: Juxtaposing Still and Dynamic Imagery. Neel Joshi, Sisil Metha, Steven Drucker, Eric Stollnitz, Hugues Hoppe, Matt Uyttendaele, and Michael F. Real-time Image-based 6-DOF Localization in Large-Scale Environments Hyon Lim, Sudipta N. Looking At You: Fused Gyro and Face Tracking for Viewing Large Imagery on Mobile Devices Neel Joshi, Abhishek Kar, Michael F. Image Based Remodeling Alex Colburn, Aseem Agarwala, Aaron Hertzmann, Brian Curless, Michael F. Video Snapshots: Creating High-Quality Images from Video Clips Kalyan Sunkavalli, Neel Joshi, Sing Bing Kang, Michael F.

My Research. Performing co-operative manipulation tasks is one of the most basic physical human-robot interaction task.

My Research

Manipulating a table collaboratively is a typical example. Traditionally, the role of the collaborating robot is fixed to a simple follower. In this research, we investigate how estimating and predicting human motion can be used by the robot to determine its role. CINT. What is CINT?

CINT

CINT is an interpreter for C and C++ code. It is useful e.g. for situations where rapid development is more important than execution time. Using an interpreter the compile and link cycle is dramatically reduced facilitating rapid development. CINT makes C/C++ programming enjoyable even for part-time programmers. CINT is written in C++ itself, with slightly less than 400,000 lines of code. Features CINT covers most of ANSI C (mostly before C99) and ISO C++ 2003.