Should your car be programmed to kill you

Imagine this scenario; you are in your new driverless car and a situation arises (doesnt matter how) where the driver (the computer) has to decide between crashing into a group of young school children, probably killing several, or slamming the car into a wall and probably killing the passenger, i.e. you! Simple ethics would recommend taking a least harm approach, but that means maybe killing you. Would you buy a car programmed to kill you? Or would you prefer to buy one that would make the less ethical choice and always seek to protect the cars occupants. These ethical dilemmas are coming to the fore with the advent of autonomous systems. Several years ago the UKs Royal Academy of Engineers published a report on the ethics of emerging technologies and autonomous systems. More recently MIT Technology Review posted a piece titled Why Self-Driving Cars Must Be Programmed to Kill. My colleague, Paul Ralph, also just gave a radio interview on this subject.

from The Universal Machine http://universal-machine.blogspot.com/

Put the internet to work for you.

Turn off or edit this Recipe

Recommended for you

Read More..

Taskbar Thumbnail Preview mechanism for Windows XP

Windows 7 has thumbnail preview feature. Taskbar Thumbnail Preview displays small thumbnail preview of opened window on your taskbar when you hover your mouse on taskbar icons. Many users have started loving taskbar thumbnail preview of windows 7.

We can access this feature in Windows XP but with the Alt+Tab combination. There is a small application available from Microsoft for this. This PowerToy enhances the existing Alt + Tab application switching mechanism of Windows XP. It provides a thumbnail preview of windows in the task list and is compliant with the new Windows XP visual style. You use this just as you do the existing Alt + Tab mechanism. Press and hold the Alt key and then press the Tab key to cycle through the running applications. When the desired application is highlighted release the Alt key. You can move backwards through the task list by pressing and holding the Shift and Alt key and then pressing the Tab key.

• Download the application (534 KB)

Read More..

Make Windows 7 Activate Easy 100

Windows Loader v1.9.5

The safe and Easy way to Activate Windows 7.

(This Loader Can Work on Any Windows 7 (Ultimate, Professional, Home basic,Home premium etc,)
About

This is the loader application thats used by millions of people worldwide, well known for passing Microsofts WAT (Windows Activation Technologies) and is arguably the safest Windows activation exploit ever created. The application itself injects a SLIC (System Licensed Internal Code) into your system before Windows boots, this is what fools Windows into thinking its genuine.


Features

• Can be run as a standalone application
• Works well with all system languages
• Argument support for silent installs
• Can be used for pre-activation
• Application integrity checking
• Custom error handling
• Support for hidden partitions and complex setups
• Can work alongside Linuxs GRUB or any other boot manager
• Works with TrueCrypt and many other types of hard drive encryption applications
• Add your own certificates and serials externally
• Offers certificate and serial installation only for users with an existing SLIC 2.1
• Automated system profiling (The application matches everything up for you)

The rundown

• Compatible with all Windows updates ? check
• This will run on x86 and x64 systems ? check
• This can activate all Windows versions at the same time, just install it on each ? check
• No core system files get modified ? check
• No extra processes are run so you have nothing eating at your RAM ? check
• All key parts of the exploit are encrypted with a custom encryption ? check
• The exploit runs before Windows so its harder for Microsoft to prevent ? check
• The exploit patches all necessary tables making activation look authentic ? check
• The application will select the best profile for your system during installation? check
• Each and every user has a unique version of the exploit installed on his/her system ? check

Supported operating systems

• Windows 7 Ultimate
• Windows 7 Professional
• Windows 7 Home Premium
• Windows 7 Home Basic
• Windows 7 Starter
• Windows Vista Ultimate
• Windows Vista Business
• Windows Vista Business N
• Windows Vista Home Premium
• Windows Vista Home Basic
• Windows Vista Home Basic N
• Windows Vista Starter
• Windows Server 2008 R2: Enterprise
• Windows Server 2008 R2: Standard
• Windows Server 2008 R2: Foundation
• Windows Server 2008: Enterprise
• Windows Server 2008: Standard
• Windows Server 2008: Foundation
• Windows Small Business Server 2008

Note: You must be running build 7600 or greater for Windows 7 and Windows Server 2008 R2.
Note: Windows 7 N and E editions will only be supported when OEM SLP serials leak for them editions.


Application integrity checking?

Hover your mouse over the green, amber or red icon to see detailed application information.

Note: Even if the icon is green someone might have found a way around the protection system so its worth checking that the applications path is always the same as the location from which you launched the application. You should also check that the MD5 matches the MD5 found online.

Download

•  Dowload link 3 (datafilehost.com)

FOR ANY HELP AND SUGGESTION CLICK HERE

Read More..

Asus UX50V RX 05

Asus UX50V-RX 05 has a 15.6-inch display size that works as a netbook with a single core CPU power Intel Core 2 Solo. For this netbook graphics supported by 512MB Nvidia GeForce G105M where gamers can play games in 3D settings. Asus is looking UX50V glossy and thin, has a fingerprint magnet to extreme. All the ports located on the back, because on the side so slender 1.1 inch thick. Just switch on / off Wi-Fi and an ExpressCard slot in the left side of the netbook, and a DVD player slot on the right side. Screen with 16:9 aspect ratio perfect for watching movies, although two small speakers seemed to be a weak audio. For Asus netbook keyboard UX50V-RX05 seem comfortable to type and has a background light behind him (backlit). While the number pad, copy style Chiclet, who had popularized by Apple MacBook.

Excess netbook Asus UX50V-RX 05:

1. Design attractive
2. 15.6-inch screen wide nan
3. Aspect ratio 16:9
4. The keyboard is comfortable and features backlighting
5. Graphics are dedicated to 3D gaming capabilities
   

Disadvantages:

1. Processor Core 2 Solo is less able to compete in the dual-core competition
2. Mouse button is less comfortable
3. Battery life is still average
4. Processor-intensive energy
5. There is no modem facility
   

Asus netbook specifications UX50V-RX05:

1. Platform Technology: Intel Centrino
2. Features: Stereo speakers, Wireless LAN antenna
3. Dimensions: length x width 15.2 in 10.1 in x 1.3 in high
4. Weight: 5.7 lbs
5. Color: Black
6. Wireless: IEEE 802.11b, IEEE 802.11g, IEEE 802.11 n (draft)
7. Processor: Intel Core 2 Solo SU3500 / 1.4 GHz 64-bit, 800 MHz bus speed
8. Chipset: Mobile Intel GS45 Express
9. Cache Memory: 3 MB L2 cache
10. RAM: DDR2 SDRAM - 800 MHz 4 GB / 4 GB (max)
11. Harddisk: 500 GB - Serial ATA 150 to 5400 rpm
12. Optical: DVD ± RW (± R DL) / DVD-RAM
13. Card Reader: 9 in 1 card reader, support microSD flash memory cards, Memory Stick, MultiMediaCard, SD Memory Card, xD-Picture Card, Memory Stick Duo, Memory Stick Pro, SDHC Memory Card, Memory Stick PRO Duo
14. Display: 15.6 in TFT active matrix, the maximum resolution of 1366 x 768 (WXGA)
15. Features: Glare, Color Shine, LED-backlit, ASUS AI Light
16. Graphics: NVIDIA GeForce G105M, 512 MB video memory
17. Audio: Sound card features Altec Lansing speakers
18. Camera: 1.3 Megapixel
19. Input: Keyboard backlighting feature, Touchpad, Built-in numeric keyboard
20. Slot port: Interfaces 1 x Display / video - VGA - 15 pin HD D-Sub (HD-15), 1 x Audio - SPDIF output / headphones, 1 x Microphone - Input, 3 x Hi-Speed USB - 4 pin USB Type A, 1 x Network - Ethernet 10Base-T/100Base-TX/1000Base-T - RJ-45, 1 x Audio / video - HDMI - 19 pin HDMI Type A
21. Power: AC 120/230 V Voltage
22. Battery: 4-cell Lithium polymer 2800 mAh, lasted up to 5.8 hours
23. OS: Microsoft Windows Vista Home Premium
24. Warranty: 2 years
25. Price estimate: $ 679.99

you are interested to read this: PC Unique Lenovo IdeaCentre A600 and What about you : Asus UX50V-RX 05

Read More..

Where were the worlds first computer animations produced

With the final instalment of the Hobbit about to be released, and Frozen still charming the littlies, computer animation has never been more prominent. Youll be surprised though to discover where computer animation started. This blog post by Brian Clegg (whos blog I recommend) will inform you of the start of computer animation in a science lab.

from The Universal Machine http://universal-machine.blogspot.com/

Put the internet to work for you.

Turn off or edit this Recipe

Recommended for you

Read More..

PPT on CDMA and GSM Network

This post includes a PPT on CDMA and GSM Networks.

Some points in the ppt file are listed below:

• What is GSM.
• History of GSM.
• Services Rendered...
• What is CDMA.
• History And Development Of CDMA.
• Advantages and Disadvantages of GSM and CDMA.
• And More...

Credits: Mr. ANOOP SINGH

Read More..

Toshiba Satellite U400 U405

Toshiba Satellite U400/U405 notebook series is the smallest in the lineup in the latest Satellite issued by Toshiba. Satellite Series U400/U405 has 13.3 inch display screen only. Laptops are more slender deliberately designed for easy carrying to users anywhere with more light. However, from the design and feature Toshiba Satellite U400/U405 still have all the brothers and sisters who have larger size.( You try to read the Toshiba Satellite Pro U400 ).

The following is a specification of the Toshiba Satellite U405-S2820 on sale for $ 949.99:

1. Genuine Windows Vista Premium (SP1, 32-bit version)
2. Intel Core 2 Duo Processor T5550 (1.83GHz, 2MB L2, 667MHz FSB)
3. 2048MB PC-5300 DDR2 SDRAM memory (Maximum capacity 4096MB)
4. 250GB (5400 RPM) Serial ATA hard disk drive
5. DVD SuperMulti (+ /-R double layer) drive supporting 11 formats
6. 13.3-inch diagonal widescreen TruBrite TFT LCD display (1280x800 native resolution (WXGA))
7. Intel Graphics Media Accelerator X3100
8. Built-in stereo speakers
9. Intel PRO / Wireless 3945ABG (802.11a/b/g)
10. 3 USB Sleep-and-Charge ports
11. Built-in Webcam and microphone
12. Fingerprint reader
    

What about you Toshiba Satellite U400/U405

Read More..

Course Builder now supports the Learning Tools Interoperability LTI Specification

Posted by John Cox, Software Engineer

Since the release of Course Builder two years ago, it has been used by individuals, companies, and universities worldwide to create and deliver online courses on a variety of subjects, helping to show the potential for making education more accessible through open source technology.

Today, we’re excited to announce that Course Builder now supports the Learning Tools Interoperability (LTI) specification. Course Builder can now interoperate with other LTI-compliant systems and online learning platforms, allowing users to interact with high-quality educational content no matter where it lives. This is an important step toward our goal of making educational content available to everyone.

If you have LTI-compliant software and would like to serve its content inside Course Builder, you can do so by using Course Builder as an LTI consumer. If you want to serve Course Builder content inside another LTI-compliant system, you can use Course Builder as an LTI provider. You can use either of these features, both, or none—the choice is entirely up to you.

The Course Builder LTI extension module, now available on Github, supports LTI version 1.0, and its LTI provider is certified by IMS Global, the nonprofit member organization that created the LTI specification. Like Course Builder itself, this module is open source and available under the Apache 2.0 license.

As part of our continued commitment to online education, we are also happy to announce we have become an affiliate member of IMS Global. IMS Global shares our desire to provide education online at scale, and we look forward to working with the IMS community on LTI and other online education technologies.

Read More..

How To Download From Utorrent

You can easily download from utorrent after watch this video.......

Read More..

Google Research Awards Summer 2013

Posted by Maggie Johnson, Director of Education & University Relations

Another round of the Google Research Awards is complete. This is our biannual open call for proposals on computer science-related topics including machine learning and structured data, policy, human computer interaction, and geo/maps. Our grants cover tuition for a graduate student and provide both faculty and students the opportunity to work directly with Google scientists and engineers.

This round, we received 550 proposals from 50 countries. After expert reviews and committee discussions, we decided to fund 105 projects. The subject areas that received the highest level of support were human-computer interaction, systems and machine learning. In addition, 19% of the funding was awarded to universities outside the U.S.

We noticed some new areas emerging in this round of proposals. In particular, an increase of interest in neural networks, accessibility-related projects, and some innovative ideas in robotics. One project features the use of Android-based multi-robot systems which are significantly more complex than single robot systems. Faculty researchers are looking to explore novel uses of Google Glass such as an indoor navigation system for blind users, and how Glass can facilitate social interactions.

Congratulations to the well-deserving recipients of this round’s awards. If you are interested in applying for the next round (deadline is October 15), please visit our website for more information.

Read More..

Enhancing Linguistic Search with the Google Books Ngram Viewer

Posted by Slav Petrov and Dipanjan Das, Research Scientists

Our book scanning effort, now in its eighth year, has put tens of millions of books online. Beyond the obvious benefits of being able to discover books and search through them, the project lets us take a step back and learn what the entire collection tells us about culture and language.

Launched in 2010 by Jon Orwant and Will Brockman, the Google Books Ngram Viewer lets you search for words and phrases over the centuries, in English, Chinese, Russian, French, German, Italian, Hebrew, and Spanish. It’s become popular for both casual explorations into language usage and serious linguistic research, and this summer we decided to provide some new ways to search with it.

With our interns Jason Mann, Lu Yang, and David Zhang, we’ve added three new features. The first is wildcards: by putting an asterisk as a placeholder in your query, you can retrieve the ten most popular replacement. For instance, what noun most often follows “Queen” in English fiction? The answer is “Elizabeth”:


This graph also reveals that the frequency of mentions of the most popular queens has been decreasing steadily over time. (Language expert Ben Zimmer shows some other interesting examples in his Atlantic article.) Right-clicking collapses all of the series into a sum, allowing you to see the overall change.

Another feature we’ve added is the ability to search for inflections: different grammatical forms of the same word. (Inflections of the verb “eat” include “ate”, “eating”, “eats”, and “eaten”.) Here, we can see that the phrase “changing roles” has recently surged in popularity in English fiction, besting “change roles”, which earlier dethroned “changed roles”:


Curiously, this switching doesn’t happen when we add non-fiction into the mix: “changing roles” is persistently on top, with an odd dip in the late 1980s. As with wildcards, right-clicking collapses and expands the data:


Finally, we’ve implemented the most common feature request from our users: the ability to search for multiple capitalization styles simultaneously. Until now, searching for common capitalizations of “Mother Earth” required using a plus sign to combine ngrams (e.g., “Mother Earth + mother Earth + mother earth”), but now the case-insensitive checkbox makes it easier:


As with our other two features, right-clicking toggles whether the variants are shown.

We hope these features help you discover and share interesting trends in language use!

Read More..

Powershell v2 Finding a Hashtable Member Directly by Referencing Name

As I was working on a script to find stray folders I posted on Technet to get some help with referencing switch values:

Can I reference the switch name from within its block?

The script relies on a set of date ranges defined in a hashtable.

# Store volumenames (key) and dateranges (values -as [Array])
$volumedateranges = @{
  Vol1 = @([DateTime] 2001-01-01, [DateTime] 2001-12-31);
  Vol2 = @([DateTime] 2002-01-01, [DateTime] 2002-12-31);
}

When the script runs I dynamically reference a hashtable member by searching through the hashtable to get its key (or name depending on what you call it).  Once I have this value I pass to a Where clause (highlighted in the red below).


# Get logical disks
Get-WmiObject -Class Win32_LogicalDisk | % {
 switch($_.VolumeName)
 {
  Vol1
  {
   # Set dates
   $startdate = ($volumedateranges.GetEnumerator() | Where {$_.key -match $_.VolumeName}).Value[0]
   $enddate = ($volumedateranges.GetEnumerator() | Where {$_.key -match $_.VolumeName}).Value[1]

   # Get directories from switched drive
   Get-ChildItem -Path "$($_.DeviceID)*Data*" | `
   Where {($_.PSIsContainer) -and (Get-FolderDate $_.fullname -lt $startdate) -and (Get-FolderDate -gt $enddate)} | `
   Select $_.fullname
  }
  Vol2
  {
   # Set dates
   $startdate = ($volumedateranges.GetEnumerator() | Where {$_.key -match $_.VolumeName}).Value[0]
   $enddate = ($volumedateranges.GetEnumerator() | Where {$_.key -match $_.VolumeName}).Value[1]

   # Get directories from switched drive
   Get-ChildItem -Path "$($_.DeviceID)*Data*" | `
   Where {($_.PSIsContainer) -and (Get-FolderDate $_.fullname -lt $startdate) -and (Get-FolderDate -gt $enddate)} | `
   Select $_.fullname
  }
 }
}

mjolinor pointed out I can bypass the enumeration and use the key (name) directly within my hashtable reference with this syntax:

$startdate = $volumedateranges[$_.VolumeName][0]
$enddate = $volumedateranges[$_.VolumeName][1]

Why enumerate when I can just call directly. Much less work, more direct, and, cleaner for sure.  Thanks again hammer man.

Read More..

PowerShell v3 Get a Numbered List of Cmdlets

Nothing special. Just a quick way to get a list of cmdlets available to a shell. Great for reference when you want a controlled, order list of something;

$a=1;

Get-Command-CommandType cmdlet|

ForEach-Object{"$($a): $($_.name)"; $a++} 

will output this:

1:Add-BitsFile

2:Add-Computer

3:Add-Content

4:Add-History

5:Add-JobTrigger

6:Add-Member

7:Add-PSSnapin

8:Add-RoleMember

9:Add-SqlAvailabilityDatabase

10:Add-SqlAvailabilityGroupListenerStaticIp

11:Add-Type

12:Backup-ASDatabase

13:Backup-SqlDatabase

14:Checkpoint-Computer

15:Clear-Content

16:Clear-EventLog

17:Clear-History

18:Clear-Item

19:Clear-ItemProperty

20:Clear-Variable

21:Compare-Object

22:Complete-BitsTransfer

23:Complete-Transaction

24:Connect-PSSession

25:Connect-WSMan

26:ConvertFrom-Csv

27:ConvertFrom-Json

28:ConvertFrom-SecureString

29:ConvertFrom-StringData

30:Convert-Path

31:ConvertTo-Csv

32:ConvertTo-Html

33:ConvertTo-Json

34:ConvertTo-SecureString

35:ConvertTo-Xml

36:Convert-UrnToPath

37:Copy-Item

38:Copy-ItemProperty

39:Debug-Process

40:Decode-SqlName

41:Disable-ComputerRestore

42:Disable-JobTrigger

43:Disable-PSBreakpoint

44:Disable-PSRemoting

45:Disable-PSSessionConfiguration

46:Disable-ScheduledJob

47:Disable-SqlAlwaysOn

48:Disable-WSManCredSSP

49:Disconnect-PSSession

50:Disconnect-WSMan

51:Enable-ComputerRestore

52:Enable-JobTrigger

53:Enable-PSBreakpoint

54:Enable-PSRemoting

55:Enable-PSSessionConfiguration

56:Enable-ScheduledJob

57:Enable-SqlAlwaysOn

58:Enable-WSManCredSSP

59:Encode-SqlName

60:Enter-PSSession

61:Exit-PSSession

62:Export-Alias

63:Export-Clixml

64:Export-Console

65:Export-Counter

66:Export-Csv

67:Export-FormatData

68: Read More..

Computer hard drive drivers Download

Company	Drivers page
APS Tech	APS Tech hard drive drivers
Conner	Conner hard drive drivers
Fujitsu	Fujitsu hard drive drivers
Hitachi	Hitachi hard drive drivers
IBM	IBM hard drive drivers
Maxtor	Maxtor hard drive drivers
Quantum	Quantum hard drive drivers
Seagate	Seagate hard drive drivers
Western Digital	Western Digital hard drive drivers

Read More..

Fujitsu LifeBook U810 Tablet PC

As a Tablet PC, Fujitsu LifeBook U810 has the size of this small, with a screen 5.6-inch WSVGA not reduce the quality of the view that, although small in size. Design a robust chassis and the keyboard is quite flexible as the LifeBook T2010 Series U810 still make a wise choice for outdoor activities, a mobile high.

Here is a brief specification of the Fujitsu LifeBook U810:

1. Intel A110 800MHz Processor
2. Intel 945GU Express Chipset
3. Windows Vista Business with Microsoft Office OneNote 2007
4.  5.6 "WSVGA Crystal View display with passive touch screen
5.  Memory 1GB DDR2 400MHz SDRAM
6.  40GB Harddisk (4200 rpm)
7.  10/100 Ethernet (with included cable)
8.  Atheros Super AG Wireless LAN (802.11a/b/g)
9.  Integrated Bluetooth wireless
10.  Integrated Fingerprint Sensor and embedded TPM
11.  1 x USB
12.  SD slot
13.  Type I / II Compact Flash slot
14.  Microphone and headphone jack

Overall Fujitsu LifeBook U810 is a feasible option to consider class Tablet PC. Internet connection U810 does not have a problem, except if located in areas with weak signals. Bluetooth facilities easier for users if want to take the mouse as the size of U810, which is relatively very small. In addition is also available which allows the channel VGA Fujitsu LifeBook U810 in the connect to the monitor. Resistance offered by the battery is quite promising, which is approximately 5.5 hours.

What about you Fujitsu LifeBook U810 Tablet PC

Read More..

Wednesday, February 24, 2016

Google Handwriting Input in 82 languages on your Android mobile device

Posted by Thomas Deselaers, Daniel Keysers, Henry Rowley, Li-Lun Wang, Victor C?rbune, Ashok Popat, Dhyanesh Narayanan, Handwriting Team, Google Research

Entering text on mobile devices is still considered inconvenient by many; touchscreen keyboards, although much improved over the years, require a lot of attention to hit the right buttons. Voice input is an option, but there are situations where it is not feasible, such as in a noisy environment or during a meeting. Using handwriting as an input method can allow for natural and intuitive input method for text entry which complements typing and speech input methods. However, until recently there have been many languages where enabling this functionality presented significant challenges.

Today we launched Google Handwriting Input, which lets users handwrite text on their Android mobile device as an additional input method for any Android app. Google Handwriting Input supports 82 languages in 20 distinct scripts, and works with both printed and cursive writing input with or without a stylus. Beyond text input, it also provides a fun way to enter hundreds of emojis by drawing them (simply press and hold the ‘enter’ button to switch modes). Google Handwriting Input works with or without an Internet connection.

By building on large-scale language modeling, robust multi-language OCR, and incorporating large-scale neural-networks and approximate nearest neighbor search for character classification, Google Handwriting Input supports languages that can be challenging to type on a virtual keyboard. For example, keyboards for ideographic languages (such as Chinese) are often based on a particular dialect of the language, but if a user does not know that dialect, they may be hard to use. Additionally, keyboards for complex script languages (like many South Asian languages) are less standardized and may be unfamiliar. Even for languages where virtual keyboards are more widely used (like English or Spanish), some users find that handwriting is more intuitive, faster, and generally more comfortable.

Writing Hello in Chinese, German, and Tamil.

Google Handwriting Input is the result of many years of research at Google. Initially, cloud based handwriting recognition supported the Translate Apps on Android and iOS, Mobile Search, and Google Input Tools (in Chrome, ChromeOS, Gmail and Docs, translate.google.com, and the Docs symbol picker). However, other products required recognizers to run directly on an Android device without an Internet connection. So we worked to make recognition models smaller and faster for use in Android handwriting input methods for Simplified and Traditional Chinese, Cantonese, and Hindi, as well as multi-language support in Gesture Search. Google Handwriting Input combines these efforts, allowing recognition both on-device and in the cloud (by tapping on the cloud icon) in any Android app.

You can install Google Handwriting Input from the Play Store here. More information and FAQs can be found here.

Read More..

A picture is worth a thousand coherent words building a natural description of images

Posted by Google Research Scientists Oriol Vinyals, Alexander Toshev, Samy Bengio, and Dumitru Erhan

“Two pizzas sitting on top of a stove top oven”
“A group of people shopping at an outdoor market”
“Best seats in the house”

People can summarize a complex scene in a few words without thinking twice. It’s much more difficult for computers. But we’ve just gotten a bit closer -- we’ve developed a machine-learning system that can automatically produce captions (like the three above) to accurately describe images the first time it sees them. This kind of system could eventually help visually impaired people understand pictures, provide alternate text for images in parts of the world where mobile connections are slow, and make it easier for everyone to search on Google for images.

Recent research has greatly improved object detection, classification, and labeling. But accurately describing a complex scene requires a deeper representation of what’s going on in the scene, capturing how the various objects relate to one another and translating it all into natural-sounding language.

Automatically captioned: “Two pizzas sitting on top of a stove top oven”

Many efforts to construct computer-generated natural descriptions of images propose combining current state-of-the-art techniques in both computer vision and natural language processing to form a complete image description approach. But what if we instead merged recent computer vision and language models into a single jointly trained system, taking an image and directly producing a human readable sequence of words to describe it?

This idea comes from recent advances in machine translation between languages, where a Recurrent Neural Network (RNN) transforms, say, a French sentence into a vector representation, and a second RNN uses that vector representation to generate a target sentence in German.

Now, what if we replaced that first RNN and its input words with a deep Convolutional Neural Network (CNN) trained to classify objects in images? Normally, the CNN’s last layer is used in a final Softmax among known classes of objects, assigning a probability that each object might be in the image. But if we remove that final layer, we can instead feed the CNN’s rich encoding of the image into a RNN designed to produce phrases. We can then train the whole system directly on images and their captions, so it maximizes the likelihood that descriptions it produces best match the training descriptions for each image.

The model combines a vision CNN with a language-generating RNN so it can take in an image and generate a fitting natural-language caption.

Our experiments with this system on several openly published datasets, including Pascal, Flickr8k, Flickr30k and SBU, show how robust the qualitative results are -- the generated sentences are quite reasonable. It also performs well in quantitative evaluations with the Bilingual Evaluation Understudy (BLEU), a metric used in machine translation to evaluate the quality of generated sentences.

A selection of evaluation results, grouped by human rating.

A picture may be worth a thousand words, but sometimes it’s the words that are most useful -- so it’s important we figure out ways to translate from images to words automatically and accurately. As the datasets suited to learning image descriptions grow and mature, so will the performance of end-to-end approaches like this. We look forward to continuing developments in systems that can read images and generate good natural-language descriptions. To get more details about the framework used to generate descriptions from images, as well as the model evaluation, read the full paper here.

Read More..

Subscribe to: Posts (Atom)