L y aura un quiz sur le vocabulaire
Download 1.72 Mb.
|
Computer vision Eye robot Poor eyesight remains one of the main obstacles to letting robots loose among humans. But it is improving, in part by aping natural vision Oct 21st 2010 The Economist ROBOTS are getting smarter and more agile all the time. They disarm bombs, fly combat missions, put together complicated machines, even play football. Why, then, one might ask, are they nowhere to be seen, beyond war zones, factories and technology fairs? One reason is that they themselves cannot see very well. And people are understandably wary of purblind contraptions bumping into them willy-nilly in the street or at home. All that a camera-equipped computer “sees” is lots of picture elements, or pixels. A pixel is merely a number reflecting how much light has hit a particular part of a sensor. The challenge has been to devise algorithms that can interpret such numbers as scenes composed of different objects in space. This comes naturally to people and, barring certain optical illusions, takes no time at all as well as precious little conscious effort. Yet emulating this feat in computers has proved tough. In natural vision, after an image is formed in the retina it is sent to an area at the back of the brain, called the visual cortex, for processing. The first nerve cells it passes through react only to simple stimuli, such as edges slanting at particular angles. They fire up other cells, further into the visual cortex, which react to simple combinations of edges, such as corners. Cells in each subsequent area discern ever more complex features, with those at the top of the hierarchy responding to general categories like animals and faces, and to entire scenes comprising assorted objects. All this takes less than a tenth of a second. The outline of this process has been known for years and in the late 1980s Yann LeCun, now at New York University, pioneered an approach to computer vision that tries to mimic the hierarchical way the visual cortex is wired. He has been tweaking his “convolutional neural networks” (ConvNets) ever since.
A ConvNet begins by swiping a number of software filters, each several pixels across, over the image, pixel by pixel. Like the brain’s primary visual cortex, these filters look for simple features such as edges. The upshot is a set of feature maps, one for each filter, showing which patches of the original image contain the sought-after element. A series of transformations is then performed on each map in order to enhance it and improve the contrast. Next, the maps are swiped again, but this time rather than stopping at each pixel, the filter takes a snapshot every few pixels. That produces a new set of maps of lower resolution. These highlight the salient features while reining in computing power. The whole process is then repeated, with several hundred filters probing for more elaborate shapes rather than just a few scouring for simple ones. The resulting array of feature maps is run through one final set of filters. These classify objects into general categories, such as pedestrians or cars. Many state-of-the-art computer-vision systems work along similar lines. The uniqueness of ConvNets lies in where they get their filters. Traditionally, these were simply plugged in one by one, in a laborious manual process that required an expert human eye to tell the machine what features to look for, in future, at each level. That made systems which relied on them good at spotting narrow classes of objects but inept at discerning anything else. Dr LeCun’s artificial visual cortex, by contrast, lights on the appropriate filters automatically as it is taught to distinguish the different types of object. When an image is fed into the unprimed system and processed, the chances are it will not, at first, be assigned to the right category. But, shown the correct answer, the system can work its way back, modifying its own parameters so that the next time it sees a similar image it will respond appropriately. After enough trial runs, typically 10,000 or more, it makes a decent fist of recognising that class of objects in unlabelled images. This still requires human input, though. The next stage is “unsupervised” learning, in which instruction is entirely absent. Instead, the system is shown lots of pictures without being told what they depict. It knows it is on to a promising filter when the output image resembles the input. In a computing sense, resemblance is gauged by the extent to which the input image can be recreated from the lower-resolution output. When it can, the filters the system had used to get there are retained. In a tribute to nature’s nous, the lowest-level filters arrived at in this unaided process are edge-seeking ones, just as in the brain. The top-level filters are sensitive to all manner of complex shapes. Caltech-101, a database routinely used for vision research, consists of some 10,000 standardised images of 101 types of just such complex shapes, including faces, cars and watches. When a ConvNet with unsupervised pre-training is shown the images from this database it can learn to recognise the categories more than 70% of the time. This is just below what top-scoring hand-engineered systems are capable of—and those tend to be much slower. This approach (which Geoffrey Hinton of the University of Toronto, a doyen of the field, has dubbed “deep learning”) need not be confined to computer-vision. In theory, it ought to work for any hierarchical system: language processing, for example. In that case individual sounds would be low-level features akin to edges, whereas the meanings of conversations would correspond to elaborate scenes. For now, though, ConvNet has proved its mettle in the visual domain. Google has been using it to blot out faces and licence plates in its Streetview application. It has also come to the attention of DARPA, the research arm of America’s Defence Department. This agency provided Dr LeCun and his team with a small roving robot which, equipped with their system, learned to detect large obstacles from afar and correct its path accordingly—a problem that lesser machines often, as it were, trip over. The scooter-sized robot was also rather good at not running into the researchers. In a selfless act of scientific bravery, they strode confidently in front of it as it rode towards them at a brisk walking pace, only to see it stop in its tracks and reverse. Such machines may not quite yet be ready to walk the streets alongside people, but the day they can is surely not far off. Science and Technology 100 UK university discoveries EurekaUK is a new report from Universities UK showcasing 50 years of research in UK universities. Here is its top 100 world-changing discoveries, innovations and research projects to come out of the UK universities in the last 50 years Wednesday July 5, 2006 Guardian Unlimited Section one: Healthy babies and birth control Producing the contraceptive pill In 1961 Herchel Smith, a researcher at the University of Manchester, developed an inexpensive way of producing chemicals that stop women ovulating during their monthly menstrual cycle. The first test tube baby Cambridge University embryologist, Robert Edwards and his colleague Patrick Steptoe, were the first to develop the IVF technique, to enable infertile women to have babies. Modern infertility treatment Medical scientists, led by Robert Winston at Imperial College London, have developed a number of tests that enable doctors to select newly created embryos that do not contain the genetic abnormalities. Scans during pregnancy: seeing babies through sound Ian Donald invented the use of ultrasound for unborn babies at the University of Glasgow 40 years ago. Babies should sleep on their backs Peter Fleming and Jem Berry at Bristol University uncovered a link between the sleeping position of babies and unexplained deaths (Sids). Spina bifida and folic acid In 1974, Nicholas Wald, then at Oxford University, discovered a way of predicting whether babies are likely to have debilitating paralysing conditions, such as spina bifida and anencephaly (where the brain is small, or missing altogether). Smoking harms babies During the 1970s Harvey Goldstein and Neville Butler, then based at the National Children's Bureau in London, studied 17,000 babies born in 1958 and discovered that babies with mothers who smoked were lower in weight by an average of 200g than other babies. Section two. Healthier and longer lives Ultrasound to detect weakened bones In the 1980s Chris Langton at Hull University was the first to develop an early detection system for osteoporosis utilising "ultrasonic" waves. Magnetic Resonance Imaging In 1976 Peter Mansfield at Nottingham University was the first to publish a successful MRI scan of a living human body part - a finger. Using light emitting chemicals to detect disease Scientists have exploited a phenomenon called "chemiluminescence" - where chemicals emit light during reactions - to develop faster and more accurate tests for allergies, anaemia, cancer and HIV. Seeing the light: inside the human body: keyhole surgery and the endoscope Harold Hopkins showed in 1954 how a bundle of tiny pin-like glass fibres allowed light and images to be transmitted along them even when they were curved- fibre optics. Pace of change; patient-controlled pacemakers Leon Abrams and Ray Lightwood at the University of Birmingham developed and implanted the first patient-controlled variable rate pacemaker. Fluoride and tooth decay Neil Jenkins, Andrew Rugg-Gunn and John Murray, based at Newcastle University, found that higher levels of fluoride in water were linked to fewer incidents of tooth decay among children. The Holy Grail of hip surgeons In 1961, John Charnley performed the first operations to replace whole hips at Wrightington hospital in Wigan. The portable defibrillator: saving lives wherever The portable defibrillator, developed in the early 1970s by Frank Pantridge at Queen's University Belfast, has saved thousands of lives. The needle-free injection In 1993 Brian Bellhouse at Oxford University invented a way of giving vaccinations and other treatments without a needle. Smoking damages your health Austin Bradford Hill and Richard Doll published a study that found that 0.5% of men with lung cancer were life-long non-smokers compared with 5% of men of the same ages in the general population. Combating a world killer: the Hepatitis B vaccine Ken Murray's search for a vaccine for Hepatitis B was prompted following a 1969 outbreak in Edinburgh that claimed 11 lives. Eradicating the Tsetse fly Scientists from the University of Greenwich have been working to eradicate the Tsetse fly from Africa through the use of a novel artificial cow. Section three: Medicine under the microscope Revealing the recipe of life James Watson and the late Francis Crick unveiled the double helix structure of the deoxyribonucleic acid or DNA on February 28, 1953. How proteins work Max Perutz pioneered the study of how proteins, the essential constituents of all living beings, work, illuminating for the first time their complex molecular structures. The building blocks of insulin In the 1950s, at the University of Cambridge, Fred Sanger revealed the exact order of the 51 basic building blocks, or amino acids, that make up the insulin molecule. The body's feel-good hormone In 1975 Hans Kosterlitz and John Hughes at Aberdeen University were the first to show that the body produces endorphins naturally. Uncovering the body's defence mechanisms In 1967 Rodney Porter at Oxford University helped to uncover the secret to the body's defence mechanisms. Genetic fingerprinting In 1985 Alec Jeffreys at the University of Leicester developed a reliable way to detect differences in the DNA of individuals - a technique now known as genetic fingerprinting. Cancer and cell division In 1987 Paul Nurse and Tim Hunt at Cancer Research UK were the first to identify the key genes that govern and regulate the cell cycle and cell division, which paved the way for progress in treating cancer. Dolly the sheep- the first cloned adult animal Ian Wilmut, a scientist at the Roslin Institute (an associated institute of the University of Edinburgh) introduced the world to Dolly the sheep, the world's first animal cloned from a cell taken from an adult animal, in 1997. Stem cells Martin Evans' early research at Cambridge University led to his discovery of embryonic stem cells - cells so early in their development that they have the potential to grow into the different cells that make up the human body. Section four: Discoveries for the digital age Fibre optics: lighting up the world In the 1950s the "founding father of fibre optics" Narinder Kapany and Harold Hopkins at Imperial College London demonstrated that light could bend, given the right encouragement. Generating information for CDs, DVDs and the internet The internet, CDs and DVDs have all been made possible through a technology called strained quantum-well lasers, first proposed by Alf Adams at Surrey University. Liquid crystal displays (LCDs) George Gray and his colleagues at Hull University first created the first stable liquid crystals for use in LCDs. Holograms Dennis Gabor at Imperial College London, invented the method of producing holograms. Manchester: birth of the first working computer Two University of Manchester scientists, Freddie Williams and Tom Kilburn, are credited with running the world's first stored program computer. The scanning electron microscope The scanning electron microscope allows researchers to peek inside materials, right down to the level of their most basic building blocks, atoms, and by so doing, to design materials that have the right properties to fit many different purposes. 3D modelling by hand "ModelMaker" is the world's first hand held 3D laser scanner that can accurately and quickly scan physical objects to make colour three-dimensional computer models. Using technology to assist disabled people Robotic Caterpillar developed by scientists at Staffordshire University allows people to perform basic day-to-day tasks by themselves. Glass, photocopiers and solar panels Nevill Mott researched into how materials conduct electricity and absorb light. Microscopic footballs Harry Kroto at Sussex University, and his US collaborators, revealed that carbon can exist as tiny spherical molecules, now known as fullerenes or buckyballs. Directory: files -> quia -> users -> johnmellor users -> Vocabulary review users -> Three Approaches to Defining Jazz Mark Gridley, Robert Maxham and Robert Hoff Copyright 1989 by Oxford University Press users -> Inventions and Inventors Web Quest Introduction users -> How do we handle these contacts? users -> A brief History of the Personal Computer users -> Lecture #13 the civil war from The Election of Lincoln to the Surrender at Appomattox Election of 1860 users -> Hours: 05 Term: Winter/Spring 2005 Days users -> Ch. 1 Thinking Geographically users -> Using a variety of sources supports the process of geographic inquiry johnmellor -> Il y aura un quiz sur le vocabulaire Download 1.72 Mb. Share with your friends:
|
