» Machine Learning

Recorded Future – CIA,Google invests into future-predicting website

Vikramaditya Jakkula — Mon, 09 Aug 2010 01:32:25 +0000

What can Recorded Future do? Lets watch:

How does it work?

Recorded Future scans Twitter accounts, blogs, and websites to find relationships, organizations, actions and incident data related to general themes.

Scour the web
Extract, rank and organize
Make it accessible and useful

It features the world’s first Temporal Analytics Engine (unlike a decision engine this uses a time series analysis). A new predictive analysis tool that allows you to visualize the future, past or present.

Prices and Plans:

About Recorded Future

Recorded Future is an early stage company headquartered in the Boston area. We have 15+ employees in various corners of the globe attacking a hard problem – organize the web in a radically new and useful way. The world’s 24×7 media flow constantly talk about time, whether it is reports of what’s transpired or statements of what’s expected to come. Recorded Future’s linguistics and statistics algorithms extract time-related information and through temporal reasoning helps users understand relationships between entities and events over time, to form the world’s first temporal analytics engine. Our customers include some of the most advanced financial institutions and leading government agencies in the world.

AI & Home

Vikramaditya Jakkula — Wed, 14 Jul 2010 07:37:12 +0000

Microsoft Research displayed some latest AI software developed by them. The focus was on AI @ home; the software displayed used different kinds of machine learning techniques, where computer software is able to learn how to carry out tasks that are useful to people without having to be programmed to perform that exact task.

[Read Silicon]

DARPA pushing machine learning into electronic warfare

Vikramaditya Jakkula — Wed, 14 Jul 2010 07:17:41 +0000

Behavioral Learning for Adaptive Electronic Warfare (BLADE) program under DARPA is developing algorithms and techniques to enable U.S. electronic warfare systems to learn to jam new RF threats automatically in the field, instead of waiting for technicians in laboratories to characterize new communications threats and develop countermeasures. This is a clear sign of advancing intelligent technology in electronic warfare.

[Read MilitaryAeroSpace & BLADE]

Watson! Dear Watson!

Vikramaditya Jakkula — Tue, 22 Jun 2010 05:06:37 +0000

For the last few years, IBM scientists have been developing the most advanced “question answering” machine, able to understand a question posed by the user, and is expected to respond with a precise answer. In other words, it must do more than what search engines like Google and Bing do, which is merely point to a set of results where you might find the answer. But Watson has to give the correct answer itself. Lets look at Watson in a trivia challenge:

[ MSNBC ]

Machine Learning Summer School – June 14,16 2010, Bangalore, India

Vikramaditya Jakkula — Fri, 11 Jun 2010 11:26:56 +0000

Machine Learning Summer School 2010’ will be hosted from June 14 – 19, 2010 at IISc Bangalore, from Yahoo! India Research & Development, in partnership with the Indian Institute of Science (IISc) Bangalore. This summer school is targeted towards academia as well as industry with a focus to deliver practical learning with or without machine learning exposure.

Schedule:

Jun-14 (morning)	Nando De Freitas/Alex Smola Introduction to ML/Graphical Models
Jun-14 (afternoon)	Nando De Freitas Gaussian Processes
Jun-15 (morning)	Chiru Bhattacharyya Support Vector Machines
Jun-15 (afternoon)	Alex Smola Graphical Models and Kernels
Jun-16 (morning)	Jayant Haritsa Association Rule Mining
Jun-16 (afternoon)	Chiru Bhattacharyya Kernel Methods
Jun-17 (morning)	Nando De Freitas Bayesian Optimization
Jun-17 (afternoon)	Jayant Haritsa Privacy Preserving Mining
Jun-18 (morning)	John Langford Transformation of learning problem
Jun-18 (afternoon)	John Langford Learning in contextual bandit settings
Jun-19 (morning)	Deepak Agarwal Recommender problems: matrix factorization
Jun-19 (afternoon)	Deepak Agarwal Recommender problems: multi-resolution models

More details visit :http://bangalore.yahoo.com/labs/summerschool.html

Microsoft Predestination: Predicting your travel destinations!

Vikramaditya Jakkula — Fri, 14 May 2010 12:53:40 +0000

Lets watch Microsoft SVP for Microsoft Research Dr. Rick Rashid talk more about Predestination:

R2 from NASA & GM Teams

Vikramaditya Jakkula — Sun, 18 Apr 2010 16:15:57 +0000

R2, short for Robonaut 2, is heading to the International Space Station aboard the space shuttle Discovery in September. R2 is able to perform experiments in micro-gravity, working with astronauts on multiple experiments.

During the last three years, a team of General Motors and NASA employees, at the Houston’s Johnson Space center, designed, engineered and build the 300-pound, human-like machine, today known as R2.

On board the ISS, R2 will be in testing phase and is said to undergo various enhancements in the future to make it truly ideal as a Space Robonaut.

[CNN]

AlSight Behavioral Analytics software wins the Security Industry Association’s New Product Showcase (NPS) Award

Vikramaditya Jakkula — Mon, 29 Mar 2010 03:51:12 +0000

Behavioral Recognition Systems, Inc. (BRS Labs) has announced that its AlSight Behavioral Analytics software has won the Security Industry Association’s New Product Showcase (NPS) Award in Video Analytics at ISC West. It is the first to use Artificial Neural Network (ANN) technology for the accurate recognition and classification of objects and behavior.

AISight Features:

Takes input from current and legacy cameras, recorded video or data files
Recognizes, remembers and tracks objects frame by frame
Rapidly learns what is acceptable behavior for any situation
Identifies and predicts abnormal, suspicious and aggressive behavior
Provides a wide range of alert systems to a human operator
Requires no additional operator training
Requires no changes to the existing infrastructure

About BRS Labs:

BRS Labs is a software development company that has created the industry’s first behavioral analysis system for video surveillance that adaptively learns behavior patterns in complex environments. BRS Labs is the only company that has been able to apply computer-vision and machine-learning capabilities to video analytics, thereby greatly enhancing operator awareness and effectiveness in improving security. No human is required to define parameters for the software to recognize behavior or objects; the software reports unusual or suspicious behaviors based on memories it has acquired through observations over time. BRS Labs was founded in November 2005 and is headquartered in Houston, Texas. The company is funded by $50 million in private equity.

[BRS Labs]

Machine Learning at Stanford University

vinayak — Sun, 28 Mar 2010 11:34:09 +0000

This time we bring to you an online video tutorial on CS 229: Machine Learning course from Stanford University. Besides 19 Lectures on machine learning you will be provided with relevant lecture notes, links other related miscellaneous stuff . Professor Andrew Ng will introduce machine learning and statistical pattern recognition which will cover topics like supervised learning, unsupervised learning, learning theory, reinforcement learning and adaptive control. Other applications of machine learning, such as to robotic control, data mining, autonomous navigation, bio informatics, speech recognition, and text and web data processing are also discussed.

This was the first lecture:- (course materials) The first 30 minutes or so of this lecture is introduction to the course and the field of machine learning in general.

In the rest of the lecture, the four main parts of the course is described in some detail along with illustrative examples :-Supervised learning: providing the algorithm a data set, supervising- learn the association between input & output, regression problems, classification problems, support vector machines- infinite number of features
Learning theory
Unsupervised learning: clustering, Cocktail party problem, independent component analysis Reinforcement learning: reward function (good dog, bad dog), feedback function

In the second lecture, Professor Andrew Ng will cover topics like linear regression, gradient descent, and normal equations and discusses how they relate to machine learning.

For more visit the home page: http://cs229.stanford.edu/

Siri…from Search to Assistance and the road ahead!

vinayak — Sun, 07 Feb 2010 16:31:25 +0000

Now that you have watched the video, you can observe that, Siri understands what you say, accomplishes tasks for you and adapts to your preferences over time. You can ask Siri questions naturally just as you will ask your assistant,like ” Book a ticket for London!” or ” Will it snow today?”. It understands information like where you are, what accounts you have on various services and the context of the last question you asked. Over time ,Siri will improve by getting to know you better and understand a broader set of tasks and soon you’ll trust it to manage many personal details in your life – from recommending a wine you might enjoy to managing your to do list.

The current version of Siri is built for iPhone 3GS and requires iPhone OS 3.1 or later. Soon, Siri will run on the iPod Touch, iPhone 3G and additional mobile platforms, as well.

Siri can get you information about Restaurants, Airline Tickets, Events, Taxis, Movies, Local and Weather Forcasts in a very simple conversation. Siri is the first Mobile Application which acts as a Virtual Personal Assistant. Its an intelligent software agent designed to have a back-and-forth conversational interaction with you as it helps you get tasks done. The creation of Siri is a joint effort of designers and engineers pooled in from various all star organisations like Google, Yahoo, Apple, Motorola, Netscape, eBay, RealTravel, SRI, NASA, and Xerox PARC.

“In terms of long-term predictions, Siri is actually an easy bet.” – The New York Times.

“It’s the closest thing to a real assistant that you can talk to that I’ve ever seen.”- Vator News

To download this iPhone application visit :

http://siri.com/download/

For more details visit :

http://siri.com/