Wednesday, April 29, 2009

Vector quantization / Another innovation from Aperio

Global Digital Pathology Leader Further Expands Its Patents Portfolio Enabling New Forms of Image Query

Vista, CA – April 28, 2009 Aperio Technologies, Inc., (Aperio), a global leader in digital pathology for the healthcare and life sciences industry, announced today that the United States Patent and Trademark Office has issued the company patent No. 7,502,519, covering systems and methods for image pattern recognition using vector quantization (VQ). This is Aperio’s second patent on the use of VQ for pattern recognition applications.
As pathology labs, hospitals, biopharma companies and educational institutions increasingly adopt digital pathology, they generate vast libraries of digital slides that play a critical role in disease management, medical research, and education. These libraries have historically been indexed for access with text-based labels such as tissue type, patient age, or primary diagnosis.
Now, Aperio’s VQ technology enables content-based image retrieval (CBIR) to allow pathologists and researchers to search libraries of digital slides using image data, and to efficiently retrieve similar images from a large image archive. The ability to search image archives using image regions of interest in addition to text-based searches represents a significant advancement in image query.
Vector quantization is a breakthrough technology providing a novel way to perform content-based image retrieval,” said Dirk Soenksen, CEO of Aperio. “The image pattern recognition technology covered by this patent is unique in that it does not rely on prior knowledge of image-based features, but involves statistical comparisons to imagery data that exhibit characteristics of interest.”
In addition to providing an efficient way to search large libraries of digital slides for image regions that match a given image, vector quantization also allows searching for exceptions, such as regions of an image which are different from previously characterized images.
Aperio’s patent portfolio encompasses all of the elements that comprise a digital pathology system, including digital slide creation, data management, advanced visualization, and image analysis. Aperio holds over 30 issued patents and pending patent applications world-wide and is the digital pathology leader in the global market with an installed base of more than 500 systems in 32 countries.

Read More

PhotoEnhancer 2.4

The latest version of PhotoEnhancer (2.4) has been released.

Version 2.4 features a 'Batch Processing' mode, for the quick enhancement of many image files just with a few clicks, as well as an improved user interface.

Read More

Tuesday, April 21, 2009

NTU talk: Mining Geotagged Photos for Semantic Understanding

Title: Mining Geotagged Photos for Semantic UnderstandingSpeaker:
Dr. Jiebo Luo, IEEE Fellow, Senior Principal Scientist with the Kodak Research Laboratories.Time: 2:30pm ~ 3:40pm, Wednesday, March 22, 2009.
Place: Room 111, CSIE building
Abstract:Semantic understanding based only on vision cues has been a challenging problem. This problem is particularly acute when the application domain is unconstrained photos available on the Internet or in personal repositories. In recent years, it has been shown that metadata captured with pictures can provide valuable contextual cues complementary to the image content and can be used to improve classification performance. With the recent geotagging phenomenon, an important piece of metadata available with many geotagged pictures is GPS information. We will describe a number of novel ways to mine GPS information in a powerful contextual inference framework that boosts the accuracy of semantic understanding. With integrated GPS-capable cameras on the horizon and geotagging on the rise, this line of research will revolutionize event recognition and media annotation.

Read More: http://nturobots.blogspot.com/2009/04/ntu-talk-mining-geotagged-photos-for.html

Web Image Retrieval ReRanking with Multi-view Clustering

General image retrieval is often carried out by a text-based search engine, such as Google Image Search. In this case, natural language queries are used as input to the search engine. Usually, the user queries are quite ambiguous and the returned results are not well-organized as the ranking often done by the popularity of an image. In order to address these problems, we propose to use both textual and visual contents of retrieved images to reRank web retrieved results. In particular, a machine learning technique, a multi-view clustering algorithm is proposed to reorganize the original results provided by the text-based search engine. Preliminary results validate the effectiveness of the proposed framework.
Read More: http://www2009.eprints.org/175/

Google Adds Search By Similarity To Image Search


Google has introduced a new experimental refinement tool that lets you click a “similar images” link beneath image search results and see a new screen of images with similar color, shape and other visual elements. The new tool has been released in Google Labs and isn’t yet available in standard image search results.
The similar images refinement is the latest in a series of tools Google has introduced to make image searching easier and more precise. These refinements include an a face filter, “exact size” filter, search suggestions and a color picker.
Image search is “harder” than text search, because true computer vision isn’t anywhere near as developed as text-based search and retrieval. The new search by similarity feature combines both analysis of images for things like color, shape, texture and so on with tagging and other techniques.
Google’s not the first to launch a similar images feature. In December last year, Microsoft added a “show similar images” capability to Live search.
Some other sites that let you search based on image similarity include Like.com, a shopping site that lets you visually compare products, Polar Rose, which detects and matches the faces in your Flickr photos and Tin Eye, a site that lets you upload an image as your query rather than typing in search terms.
Want to know more about computer based image retrieval? See my Teaching Google To See Images from April of last year.

Article from:http://searchengineland.com/google-adds-search-by-similarity-to-image-search-17764

Monday, April 13, 2009

The 2009 International Workshop on 3-D Digital Imaging and Modeling

The field of 3D imaging, modeling and visualization has undergone a rapid growth over the last decade. While some research issues have received significant attention and matured into stable solutions, new problems and goals are emerging as the focus of the 3-D research community. Improved methods for the acquisition of 3-D information by optical means are driven by new algorithmic approaches in computer vision and image processing.

Advanced methods for the processing and transformation of geometric information open new types of applications. As part of ICCV 2009, the 3DIM 2009 Workshop will bring together researchers interested in all aspects of the acquisition, processing and modeling of 3-D information and their applications.

The 3DIM 2009 Committee invites you to submit high quality original full papers by *JUNE 01, 2009*.

PLEASE NOTE THE EXTENDED SUBMISSION DEADLINE AND PUBLICATION CALENDAR.

The papers will be reviewed by an international Program Committee (see web site). Accepted papers will be presented in single-track oral sessions as well as a poster session (all papers are allocated the same number of pages).

The Workshop Proceedings will be published by the IEEE Computer Society Press as part of the ICCV 2009 Proceedings, and archived in their Digital Library.

Full details on the paper format, electronic submission procedure and conference venue are available on the 3DIM 2009 Workshop web site.

See web site http://www.3DIMconference.org for complete and updated information.

October 3-4, 2009
Kyoto, Japan

Paper submission due: June 01, 2009
Notification of results: July 15, 2009
Final paper due: August 14, 2009

Exalead : A Look Into Semantic Image Search

Article from:http://www.searchenginejournal.com/exalead-a-look-into-semantic-image-search/5283/

Exalead, the search engine founded by Francois Bourdoncle (earlier involved with AltaVista) has a number of features that makes it a bit of everything. It has vertical search option (including blog, video, image search), products tailored for enterprises and the search engine sports nifty features such as – Truncation, Thumbnails of pages and more. But semantics is my theme and this article focuses on Exalead’s image search.

Unwrapping the Exalead Image Search

By leveraging the core image recognition engine from LTU Technologies Exalead offers image-searching features that can be used for comparison and classification of images.

The image recognition engine consists of two modules:

1. Image DNA generator (Image Analysis): The engine generates a numerical vector (DNA) encoding image information such as color, texture, shape, spatial configuration, image quality, image size, image brightness, contrast, distortion, object translation, object rotation and scale.

2. Semantic description Generator (Image Description): The DNA for the image is classified on the basis pattern recognition vis-à-vis a knowledge base using state of the art techniques modeled after behavior of human subjects.

The analyzer and describer work in tandem in real-time and also include “learning” capabilities to enhance the search experience. More information on the technology is available at the official LTU website.

This classification scheme fits well with the surplus options that Exalead provides for narrowing down image search. You have the option to search images based on size (Small, Medium, Large), content (Face), pixel size (800×600, 1024×768, 1280×768) and image color (color, Gray scale, Black & White).

Read More

9 Semantic Search Engines That Will Change the World of Search

Article from:http://www.facebook.com/ext/share.php?sid=157851995690&h=2ehKk&u=ojOUW&ref=nf

The ideal search engine would be able to match the search queries to the exact context and return results within that context. While Google, Yahoo and Live continue to hold sway in search, here are the engines that take a semantics (meaning) based approach, the end result being more relevant search results which are based on the semantics and meaning of the query, and not dependent upon preset keyword groupings or inbound link measurement algorithms, which make the more traditional search engines easier to game, thus including more spam oriented results.

Here is a wrap up of some of the top semantic search engines which we’ve covered previously, and some updates on their research.

1. Hakia

The brainchild of Dr. Riza C. Berkan, tries to anticipate the questions that could be asked relating to a document and uses them as the gateways to the content.

The search queries are mapped to the results and ranked using an algorithm that scores them on sentence analysis and how closely they match the concept related to the query.

Hakia semantic search is essentially built around three evolving technologies:

  1. OntoSem (sense repository)
  2. QDEX (Query indexing technique)
  3. SemanticRank algorithm
  • OntoSem is Hakia’s repository of concept relations, in other words, a linguistic database where words are categorized into the various “senses” they convey.
  • QDEX is Hakia’s replacement for the inverted index that most engines use to save web content. QDEX extracts all possible queries relating to the content (leveraging the OntoSem for meaning) and these become the gateways to the original document. This process greatly reduces the data set that the indexer has to deal with while querying data on-the-fly. An advantage when you considering the wide swath of data the engine would have to search if it were an inverted index.
  • Finally, the SemanticRank algorithm independently ranks content on the basis of more sentence analysis. Credibility and age of the content is also used to determine relevancy.

Hakia performs pure analysis of content irrespective of links or clickthroughs among the documents (they are opposed to statistical models for determining relevance).

The engine has also started using the Yahoo BOSS service and also presents results in a “gallery” with categories for different content matching the query. Users can also request to try out the the incremental changes that are being tried at Hakia’s Lab.

2. Kosmix

The search company has takes its categorization concept further by providing users with a dashboard of content, aptly called - ” Your guide to the Web”. The company’s focus on informational search makes it suitable for topics when you want information on it rather than look for a particular answer or URL. For example, the search for Credit Default Swap provided a great mix of links, videos and tweets to get me started. Kosmix received $20 million of funding from Time Warner in late 2008. Its content aggregating technology will become more important as content on the web grows.

Read More

Sunday, April 12, 2009

Virtual Participation in WMSCI 2009 is Now Possible

Deadlines for Virtual Participation in The 13th World Multi-Conference on Systemics, Cybernetics and Informatics: WMSCI 2009 (http://www.ICTconfer.org/wmsci) (Orlando, Florida, USA. July 10th-13th, 2009)

Extended Abstracts submissions and Invited Virtual Sessions proposals: May 6th, 2009 Authors notifications: June 1st, 2009 Camera ready, full papers: June 22nd, 2009

Submissions made for Virtual Participation will go trough the same reviewing processes of the regular papers (double-blind, non-blind, and participative peer reviewing) and, if accepted (according to the same acceptance policy), they will be included in the proceedings and will be eligible for journal publication if they are, according to their reviewers, among the best 10%-20% of those physically and virtually presented at the conference.

Each regular session, included in the conference program, will be associated to a corresponding virtual session where all final versions of the articles to be presented will be displayed and authors can comment them via electronic forums. Registered authors of virtual participations will have access to all conference program sessions via their respective associated virtual sessions. Their article will be displayed like the regular ones. Virtual authors also have the option of sending, besides the final version of their article in a PDF document, an electronic presentation (PowerPoint, flash, etc. and/or a 15-20 minutes video).

After paying the respective shipping and handling costs, registered authors of virtual participation, who paid their registration fee, can get delivered the same conference material that the regular attendees receive at the registration desk.

Authors of accepted papers who registered in the conference will have access to the reviews made to their submission so they can accordingly improve the final version of their papers. Non-registered authors may not have access to the reviews of their respective submissions.

Awards will be granted to the best paper of those presented at each session. From these sessions' best papers, the best 10%-20% of the papers presented at the conference will be selected for their publication in Volume 7 of JSCI Journal (http://www.j-sci.com/Journal/SCI/). Libraries of journal authors' organizations will receive complimentary subscriptions of at least one volume (6 issues).

Saturday, April 11, 2009

LEEGLE - A challenge in object class recognition in sequences

LEEGLE is a challenge in object recognition and categorization initiated and organized by University of Ljubljana, Visual Cognitive Systems Laboratory, and mainly intended for students of computer vision. The challenge is organized twice a year, at the end of each semester. It is an opportunity for students to test their knowledge of computer vision and compete with teams from other universities.

Three months prior to the competition the training and validation data is made available. The training data consists of a set of 3D views of various toy objects pertaining to several object classes. A validation sequence simulates a trajectory of a camera through a toy world, depicting objects with high variations in scale, poses, orientation, lighting conditions, and some objects are likely to be occluded. The task is to detect, recognize and categorize as many of the known objects as possible. For the competition a test image sequence is given, where the algorithms developed are to be evaluated within 48 hours. Participating groups should submit their results, and resulting scores will be available on the web site.

The test sequence contains many objects in various poses, scales and orientations. The goal of the challenge is to detect, recognize and categorize as many of them as possible. For each challenge, a training set of images and a validation set with ground truth annotations is made available.

Once the test image sequence is given, the algorithms developed should be evaluated within 48 hours. For each test image the set of bounding boxes around the detected objects with the corresponding class labels should be submitted. The bounding box is determined by the upper left coordinate (X1, Y1) and the lower right coordinate (X2, Y2), where (0, 0) represents the upper left corner of the image. A detection will be counted as correct if the bounding box overlaps with the ground truth more than 40% and vice versa, and has a correct label. Since one object can belong to several categories, each correctly assigned class will be counted as a true positive (TP) and each wrong label will mean a false positive (FP). An exception is the 'zombie' class which should be avoided (it is a distractor object and should not be labeled). The ground truth bounding boxes outline only the visible part of the object. Similarly, your results shall report bounding boxes of the visibe part of the detected object (and not the dimensions inferred from the training set images), omitting thus the parts clipped by the image border or by other objects in the scene.

An example of an annotated scene image (ground truth):

The images on the simulated path will be captured from viewpoints with a constant height and tilt, both approximately the same as in the training sequence. The illumination, backgrounds, and the configuration of the scene will vary. Objects can be augmented with occluding parts (e.g. carrying tools), but the pose will be approximately the same as in the training set.

http://vicos.fri.uni-lj.si/leegle/

Thursday, April 9, 2009

6th European Lisp Workshop

This year, and for the first time, the workshop proceedings will be published in the ACM Digital Library. Also, the workshop will feature interactive tutorial/demo/coding sessions (see below).

Overview

"...Please don't assume Lisp is only useful for Animation and Graphics, AI, Bio-informatics, B2B and E-Commerce, Data Mining, EDA/Semiconductor applications, Expert Systems, Finance, Intelligent Agents, Knowledge Management, Mechanical CAD, Modeling and Simulation, Natural Language, Optimization, Research, Risk Analysis, Scheduling, Telecom, and Web Authoring just because these are the only things they happened to list."

-- Kent Pitman

Lisp, one of the eldest computer languages still in use today, is gaining momentum again. The structure of Lisp makes it easy to extend the language or even to implement entirely new dialects without starting from scratch, making it the ideal candidate for writing Domain Specific Languages. Common Lisp, with the Common Lisp Object System (CLOS), was the first object-oriented programming language to receive an ANSI standard and retains the most complete and advanced object system of any programming language, while influencing many other object-oriented programming languages that followed.

This workshop will address the near-future role of Lisp-based languages in research, industry and education. We solicit contributions that discuss the opportunities Lisp provides to capture and enhance the possibilities in software engineering. We want to promote lively discussion between researchers proposing new approaches and practitioners reporting on their experience with the strengths and limitations of current Lisp technologies.

The workshop will have two components: there will be formal talks, and interactive turorial/demo/coding sessions.

Papers

Formal presentations in the workshop should take between 20 minutes and half an hour; additional time will be given for questions and answers. Suggested topics include (but are not limited to):

  1. - Experience reports / Case studies
  2. - Educational approaches
  3. - Software Evolution
  4. - Development Aids
  5. - Persistent Systems
  6. - Dynamic Optimization
  7. - Implementation techniques
  8. - Hardware Support
  9. - Efficiency / Distribution / Parallel programming
  10. - Macro-, reflective-, meta- and/or rule-based development approaches
  11. - Protocol Meta-programming and Libraries
  12. - Context-Oriented, Domain-Oriented and Generative Programming

The Incredible Convenience of Mathematica Image Processing

Article from Theodore Gray, Co-founder, Director of User Interfaces
Full Article: http://blog.wolfram.com/2008/12/01/the-incredible-convenience-of-mathematica-image-processing/

Here’s a clown fish, and a command that breaks it into 40-pixel squares.

Fish image broken into 40-pixel squares

By the way, we’re seeing another neat thing about the integration of images with Mathematica’s typeset input/output system. This result isn’t an image, it’s a list of images. Lists are general things in Mathematica, and lists of images are no exception. For example, here are the image patches in reverse order.

Fish image's patches in reverse order

And here they are sorted by average pixel value (roughly by brightness):

Fish image's patches sorted by average pixel value

And here are the images sorted into a scatter plot where the x axis represents the red component, the y axis represents the green component, and the size represents the blue component. (Image processing commands are fully integrated and compatible with charting commands. This is, after all, Mathematica, where one comes to expect things to work with each other.)

Image patches in a scatter plot

See how all the green patches are smooth and all the red patches are high-contrast? It might be interesting to look at these patches in a similarity network rather than a scatter plot. Here’s some code to do that.

Coding a function to make similarity graphs

This function identifies patches that are similar in color, then connects them into a network. The parameter says how many neighbors to look at before building the network.

Building a three-neighbor similarity network

Remember how we made a Manipulate to play with the contrast of an image? How about a Manipulate to play with the number of neighbors?

Manipulating the number of neighbors in our similarity graph

QDA Miner 3.2

QDA Miner is an easy-to-use qualitative data analysis software package for coding textual data, annotating, retrieving and reviewing coded data and documents. The program can manage complex projects involving large numbers of documents combined with numerical and categorical information. QDA Miner also provides a wide range of exploratory tools to identify patterns in codings and relationships between assigned codes and other numerical or categorical properties.

This new version introduces several useful features, including:

  • Virtual grouping of codes for flexible analysis of hierarchical coding.
  • New bubble charts and stacked bar charts for displaying relationships between codings and variables.
  • Improved data management features (e.g., incremental importation).
  • Integrated statistical routines for assessing the uni-variate and bi-variate distributions of numerical and categorical variables.
  • New intuitive text-retrieval tools for easier data inspection.
  • Improved integration with WordStat for in-depth analysis of query results.
  • Several speed optimizations for large projects involving numerous cases as well as very large documents.

More detailed information on these new features can be found at:

http://www.kovcomp.com/QDAMiner/QDA3new.html

A full description of the program is at:

http://www.kovcomp.com/QDAMiner/qdambroc.html

A demo version of the program can be downloaded at the site listed below. Existing users of QDA Miner v3.0 or v3.1 can upgrade for free by installing this demo version.

http://www.kovcomp.com/QDAMiner/downl.html

International Symposium on Optical Engineering and Photonic Technology: OEPT 2009

Last extension of deadlines of The International Symposium on Optical Engineering and Photonic Technology: OEPT 2009

(http://www.ICTconfer.org/oept)

(Orlando, Florida, USA. July 10th-13th, 2009)

Deadlines for both, Regular Face-to-Face and Virtual Participation

Papers/abstracts submissions and Invited Sessions Proposals: May 6th, 2009 Authors Notification: June 1st, 2009 Camera ready, full papers: June 22nd, 2009

Submissions made for Virtual Participation will go trough the same reviewing processes of the regular papers (double-blind, non-blind, and participative peer reviewing) and, if accepted (according to the same acceptance policy), they will be included in the proceedings and will be eligible for journal publication if they are, according to their reviewers, among the best 10%-20% of those physically and virtually presented at the conference.

Each regular session, included in the conference program, will be associated to a corresponding virtual session where all final versions of the articles to be presented will be displayed and authors can comment them via electronic forums. Registered authors of virtual participations will have access to all conference program sessions via their respective associated virtual sessions. Their article will be displayed as the regular ones. Virtual authors also have the option of sending, besides the final version of their article in a PDF document, an electronic presentation (PowerPoint, flash, etc. and/or a 15-20 minutes video).

After paying the respective shipping and handling costs, registered authors of virtual participation, who paid their registration fee, can get delivered the same conference material that the regular attendees receive at the registration desk.

Authors of accepted papers who registered in the conference will have access to the reviews made to their submission so they can accordingly improve the final version of their papers. Non-registered authors may not have access to the reviews of their respective submissions.

Awards will be granted to the best paper of those presented at each session. From these session's best papers, the best 10%-20% of the papers presented at the conference will be selected for their publication in Volume 7 of JSCI Journal (http://www.j-sci.com/Journal/SCI/). Libraries of journal author's organizations will receive complimentary subscriptions of at least one volume (6 issues).

PCI 2009 - 13th Panhellenic Conference on Informatics

10 - 12 September 2009 Corfu, Greece

The Greek Computer Society (ΕΠΥ), the Ionian University, Department of Informatics and University of Piraeus, Department of Informatics organize the 13th Panhellenic Conference on Informatics (PCI 2009) at Corfu Island, Greece, during 10 - 12 of September, 2009.

The PCI is an event established by the Greek Computer Society. The 1st Conference took place at Athens (1984), the 2nd at Thessaloniki (1988), the 3rd at Athens (1991), the 4th at Patras (1993), the 5th at Athens (1995), the 6th at Athens (1997), the 7th at Ioannina (1999), the 8th at Nicosia Cyprus (2001), the 9th at Thessaloniki (2003), the 10th at Volos (2005), the 11th at Patras (2007) and the 12th at Samos (2008). This year PCI 2009 will take place at Corfu Island.

The PCI 2009 Conference will run in parallel sessions, with invited talks, research and case study tracks. Authors are invited to submit papers in any area of Informatics, Computer Science, Computer Engineering, Telecommunications, and Information Systems. Topics of interest include, but are not limited to, the following:

  • Algorithms and Data Structures
  • Artificial Intelligence
  • Bioinformatics
  • Business Intelligence
  • Communication and Information Systems Security
  • Computational Science
  • Computer and Communication Networks
  • CRM and ERP systems
  • Cultural and Museum Information Systems
  • Databases
  • Data Mining
  • Digital Libraries
  • eCommerce, eBusiness, eGovernment, eHealth
  • Education Technologies
  • Graphics, Visualization, Multimedia and Virtual Reality
  • Grid and Cluster Computing
  • Hardware and Architecture
  • Human-Computer Interaction
  • Image and Video Processing
  • Information Retrieval

Wednesday, April 8, 2009

img(Rummager) Server Side is ready

Combine low level features with keywords. The server side part of the Img(Rummager) undertakes the execution of image retrieval based on Flickr keywords (tags) and creates XML index files containing the descriptors of these images.

When the user of the img(Rummager) client version selects the application server search, the img(Rummager) automatically connects to its server version and downloads an XML file with the available keywords. Thus far the application supports 100 keywords. The options appear in a drop down menu.

The user selects the desired keyword, preferred descriptor for the search, and imports a query image. The application downloads a second XML file containing the descriptors for 500 images tagged with the keyword specified by the user. It should be noted that images on Flickr are tagged by the users who have uploaded them. No image is stored on the user’s computer during the retrieval process.

If you are using version 2009-3-2 1.0.334+ of img(Rummager) you don't have to do anything. New features will be automatically appear in the application.

Here are some screenshots from the server side part of the img(Rummager) in action. Very special thanks to my students (and friends): Congeo, Eirini Crouse and to my colleague Kostantinos Ioannidis

DSC00391  

New Folder

Developing a Document Image Retrieval System

Greetings, my name is Konstantinos Zagoris and I am a close friend of Savvas Chatzichristofis and the developer of the img(Anaktisi). I have been invited to this blog to describe a field of Retrieval Information: the Image Document Retrieval System (DIRS) through word spotting.

This technique performs the word matching directly in the document images bypassing OCR and using word-images as queries. The entire system consists of the Offline and the Online procedures. In the Offline procedure, the document images are analyzed and the results are stored in a database. Three main stages, the preprocessing, the word segmentation and the feature extraction stages, constitute the offline procedure. A set of features, capable of capturing the word shape and discard detailed differences due to noise or font differences are used for the word-matching process. The Online procedure consists of four components: the creation of the query image, the preprocessing stage, the feature extraction stage, and finally, the matching procedure.

The overall structure of the Document Image Retrieval System.

In contrast to the descriptors that they hosted in img(Anaktisi), this descriptor uses primarily shape features. The image below depicts the descriptor and the features that it contains. These features was selected in such way that describe satisfactorily the shape of the query words while at the same moment they suppress small differences due to noise, size and type of fonts.

zag_fig2

The description of the above features can be found in the journal article:

or in the conference paper (in a compact form):

You can find the presentation of the above conference paper here. Below is a more simple version for web presentations purposes.

A very early (and rough) version of the proposed DIRS is described in the conference paper:

The proposed system is implemented with the help of the Visual Studio 2008 and is based on the Microsoft .NET Framework 3.5. The programming language which is used is the C#. For user interaction the application employs the AJAX/Javascript and HTML technologies.

The image documents included in the database are created artificially from various texts and then noise was added in order to implement in parallel a text search engine which makes easier the verification and evaluation of the search results of the DIRS system. Furthermore, the database used by the implemented DIRS is the Microsoft SQL Server 2005.

clip_image002

The web address of the implemented system is the http://orpheus.ee.duth.gr/irs2_5

The advantage of the described method is the resilience to the noise. An example of a noisy document is depicted in the below image. This document is the retrieval result for the word “literature”.

zag_fig3

Read Part II

For more information or questions email me at kzagoris@gmail.com.

Dr Konstantinos Zagoris (http://www.zagoris.gr) received the Diploma in Electrical and Computer Engineering in 2003 from Democritus University of Thrace, Greece and his phD from the same univercity in 2010. His research interests include document image retrieval, color image processing and analysis, document analysis, pattern recognition, databases and operating systems. He is a member of the Technical Chamber of Greece.

Wednesday, April 1, 2009

Cognitive Autoheuristic Distributed-Intelligence Entity

Introducing CADIE

Research group switches on world's first "artificial intelligence" tasked-array system.

For several years now a small research group has been working on some challenging problems in the areas of neural networking, natural language and autonomous problem-solving. Last fall this group achieved a significant breakthrough: a powerful new technique for solving reinforcement learning problems, resulting in the first functional global-scale neuro-evolutionary learning cluster.

Since then progress has been rapid, and tonight we're pleased to announce that just moments ago, the world's first Cognitive Autoheuristic Distributed-Intelligence Entity (CADIE) was switched on and began performing some initial functions. It's an exciting moment that we're determined to build upon by coming to understand more fully what CADIE's emergence might mean, for Google and for our users. So although CADIE technology will be rolled out with the caution befitting any advance of this magnitude, in the months to come users can expect to notice her influence on various google.com properties. Earlier today, for instance, CADIE deduced from a quick scan of the visual segment of the social web a set of online design principles from which she derived this intriguing homepage.

These are merely the first steps onto what will doubtless prove a long and difficult road. Considerable bugs remain in CADIE'S programming, and considerable development clearly is called for. But we can't imagine a more important journey for Google to have undertaken.

For more information about CADIE see this monograph, and follow CADIE's progress via her YouTube channel and blog.