(Press-News.org) In today's digitally driven world, access to information appears limitless.
But when you have something specific in mind that you don't know, like the name of that niche kitchen tool you saw at a friend's house, it can be surprisingly hard to sift through the volume of information online and know how to search for it. Or, the opposite problem can occur – we can look up anything on the Internet, but how can we be sure we are finding everything about the topic without spending hours in front of the computer?
Computer scientists from the University of Washington and the Allen Institute for Artificial Intelligence in Seattle have created the first fully automated computer program that teaches everything there is to know about any visual concept. Called Learning Everything about Anything, or LEVAN, the program searches millions of books and images on the Web to learn all possible variations of a concept, then displays the results to users as a comprehensive, browsable list of images, helping them explore and understand topics quickly in great detail.
"It is all about discovering associations between textual and visual data," said Ali Farhadi, a UW assistant professor of computer science and engineering. "The program learns to tightly couple rich sets of phrases with pixels in images. This means that it can recognize instances of specific concepts when it sees them."
The research team will present the project and a related paper this month at the Computer Vision and Pattern Recognition annual conference in Columbus, Ohio.
The program learns which terms are relevant by looking at the content of the images found on the Web and identifying characteristic patterns across them using object recognition algorithms. It's different from online image libraries because it draws upon a rich set of phrases to understand and tag photos by their content and pixel arrangements, not simply by words displayed in captions.
Users can browse the existing library of roughly 175 concepts. Existing concepts range from "airline" to "window," and include "beautiful," "breakfast," "shiny," "cancer," "innovation," "skateboarding," "robot," and the researchers' first-ever input, "horse."
If the concept you're looking for doesn't exist, you can submit any search term and the program will automatically begin generating an exhaustive list of subcategory images that relate to that concept. For example, a search for "dog" brings up the obvious collection of subcategories: Photos of "Chihuahua dog," "black dog," "swimming dog," "scruffy dog," "greyhound dog." But also "dog nose," "dog bowl," "sad dog," "ugliest dog," "hot dog" and even "down dog," as in the yoga pose.
The technique works by searching the text from millions of books written in English and available on Google Books, scouring for every occurrence of the concept in the entire digital library. Then, an algorithm filters out words that aren't visual. For example, with the concept "horse," the algorithm would keep phrases such as "jumping horse," "eating horse" and "barrel horse," but would exclude non-visual phrases such as "my horse" and "last horse."
Once it has learned which phrases are relevant, the program does an image search on the Web, looking for uniformity in appearance among the photos retrieved. When the program is trained to find relevant images of, say, "jumping horse," it then recognizes all images associated with this phrase.
"Major information resources such as dictionaries and encyclopedias are moving toward the direction of showing users visual information because it is easier to comprehend and much faster to browse through concepts. However, they have limited coverage as they are often manually curated. The new program needs no human supervision, and thus can automatically learn the visual knowledge for any concept," said Santosh Divvala, a research scientist at the Allen Institute for Artificial Intelligence and an affiliate scientist at UW in computer science and engineering.
The research team also includes Carlos Guestrin, a UW professor of computer science and engineering. The researchers launched the program in March with only a handful of concepts and have watched it grow since then to tag more than 13 million images with 65,000 different phrases.
Right now, the program is limited in how fast it can learn about a concept because of the computational power it takes to process each query, up to 12 hours for some broad concepts. The researchers are working on increasing the processing speed and capabilities.
The team wants the open-source program to be both an educational tool as well as an information bank for researchers in the computer vision community. The team also hopes to offer a smartphone app that can run the program to automatically parse out and categorize photos.
INFORMATION:
This research was funded by the U.S. Office of Naval Research, the National Science Foundation and the UW.
For more information, contact Farhadi at ali@cs.washington.edu or 206-221-8976 and Divvala at santosh@cs.washington.edu.
Posted with photo, video: http://www.washington.edu/news/2014/06/12/new-computer-program-aims-to-teach-itself-everything-about-anything/
New computer program aims to teach itself everything about anything
2014-06-12
[Attachments] See images for this press release:
ELSE PRESS RELEASES FROM THIS DATE:
Grit better than GRE at predicting success in STEM fields
2014-06-12
Selecting graduate students in the fields of science and engineering based on an assessment of their character instead of relying almost entirely on their scores on a standardized test would significantly improve the quality of the students that are admitted and, at the same time, boost the participation of women and minorities in these key disciplines.
That is the argument made in the essay "A test that fails" published in the June 12 issue of the journal Nature. The authors are Associate Professor of Physics Casey Miller of the University of South Florida and Keivan ...
Researchers uncover new insights into developing rapid-acting antidepressant for treatment-resistant depression
2014-06-12
DALLAS – June 12, 2014 – UT Southwestern Medical Center researchers have generated fresh insights that could aid in the development of rapid-acting antidepressants for treatment-resistant depression.
The researchers found that by blocking NMDA receptors with the drug ketamine, they could elicit rapid antidepressant effects in patients with treatment-resistant depression. Ketamine was developed as an anesthetic, but is better known publicly for its abuse as the party drug Special K. Researchers are now seeking alternatives because ketamine can produce side effects that ...
Good bacteria armed with antibiotic resistance protect gut microbiome
2014-06-12
Researchers from Case Western Reserve University in Cleveland have discovered that populating the gastrointestinal (GI) tracts of mice with Bacteroides species producing a specific enzyme helps protect the good commensal bacteria from the harmful effects of antibiotics. Their research is published ahead of print in Antimicrobial Agents and Chemotherapy.
Antibiotics are powerful weapons against pathogens, but most are relatively indiscriminate, killing the good bacteria, along with the bad. Thus, they may render patients vulnerable to invasion, particularly by virulent, ...
Neural reward response may demonstrate why quitting smoking is harder for some
2014-06-12
For some cigarette smokers, strategies to aid quitting work well, while for many others no method seems to work. Researchers have now identified an aspect of brain activity that helps to predict the effectiveness of a reward-based strategy as motivation to quit smoking.
The researchers observed the brains of nicotine-deprived smokers with functional magnetic resonance imaging (fMRI) and found that those who exhibited the weakest response to rewards were also the least willing to refrain from smoking, even when offered money to do so.
"We believe that our findings may ...
Proteins causing daytime sleepiness tied to bone formation, target for osteoporosis
2014-06-12
DALLAS – June 12, 2014 – Orexin proteins, which are blamed for spontaneous daytime sleepiness, also play a crucial role in bone formation, according to findings by UT Southwestern Medical Center researchers. The findings could potentially give rise to new treatments for osteoporosis, the researchers say.
Orexins are a type of protein used by nerve cells to communicate with each other. Since their discovery at UT Southwestern more than 15 years ago, they have been found to regulate a number of behaviors, including arousal, appetite, reward, energy expenditure, and wakefulness. ...
Long-range tunneling of quantum particles
2014-06-12
This news release is available in German. One of the most remarkable consequences of the rules in quantum mechanics is the capability of a quantum particle to penetrate through a potential barrier even though its energy would not allow for the corresponding classical trajectory. This is known as the quantum tunnel effect and manifests itself in a multitude of well-known phenomena. For example, it explains nuclear radioactive decay, fusion reactions in the interior of stars, and electron transport through quantum dots. Tunneling also is at the heart of many technical ...
Vast genetic diversity among Mexicans found in large-scale study
2014-06-12
The first large-scale, comprehensive analysis of the genomic diversity of Mexico — led by researchers at the Stanford University School of Medicine, the University of California-San Francisco and the Mexican National Institute of Genomic Medicine — has identified a dazzling mosaic of genotypes and population substructures across the country.
Some groups are as genetically different from one another as Europeans are from East Asians.
The study, which will be published June 13 in Science, soundly refutes the current practice of lumping together Mexicans or Latinos as ...
Broad Institute, MGH researchers chart cellular complexity of brain tumors
2014-06-12
Scientists from the Broad Institute and Massachusetts General Hospital (MGH) have conducted a first-of-its-kind study that characterizes the cellular diversity within glioblastoma tumors from patients. The study, which looked at the expression of thousands of genes in individual cells from patient tumors, revealed that the cellular makeup of each tumor is more heterogeneous than previously suspected. The findings, which appear online in Science Express, will help guide future investigations into potential treatments for this devastating disease.
This is the first time ...
Mexican genetics study reveals huge variation in ancestry
2014-06-12
In the most comprehensive genetic study of the Mexican population to date, researchers from UC San Francisco and Stanford University, along with Mexico's National Institute of Genomic Medicine (INMEGEN), have identified tremendous genetic diversity, reflecting thousands of years of separation among local populations and shedding light on a range of confounding aspects of Latino health.
The study, which documented nearly 1 million genetic variants among more than 1,000 individuals, unveiled genetic differences as extensive as the variations between some Europeans and Asians, ...
Father's age influences rate of evolution
2014-06-12
The offspring of chimpanzees inherit 90% of new mutations from their father, and just 10% from their mother, a finding which demonstrates how mutation differs between humans and our closest living relatives, and emphasises the importance of father's age on evolution.
Published today in Science, researchers from the Wellcome Trust Centre for Human Genetics and the Biomedical Primate research Centre in the Netherlands looked at whether, in chimpanzees, there was a heightened risk of fathers passing on mutations to their children compared to humans.
In humans, each individual ...