Feed aggregator

Book Review: Build Your Own Website: A Comic Guide to HTML, CSS, and WordPress

Slashdot - Wed, 17/12/2014 - 8:58pm
MassDosage writes "At the the risk of exposing my age I remember building my first website using a rudimentary Unix text editor (Joe) and carefully handcrafting the Hypertext Markup Language (HTML) while directly logged on to the web server it was being served from. Back then Cascading Style Sheets (CSS) weren't even a glint in the eyes of their creators. A lot has changed and there's now a world of fancy WYSIWYG web page editors to choose from as well as Content Management Systems that allow you to create websites without looking at the underlying code at all. While this is all very useful and allows less technical people to create websites I still feel that having at least some knowledge of how everything works under the hood is empowering — especially in situations where you want to go beyond the limits placed on you by a certain tool. This is where Build Your Own Website: A comic guide to HTML, CSS and Wordpress comes into the picture. Its aim is to enable people new to web development to learn the subject by teaching the fundamentals of HTML and CSS first and only then describing how to use a Content Management System (CMS) — in this case Wordpress. While Wordpress might not be everyone's kettle of fish it's a good choice as an example of a modern CMS that is easily accessible and very popular. The concepts presented are simple enough that it should be easy enough for a reader to apply them to a different CMS should they want to. Read below for The rest of MassDosage's review.

Read more of this story at Slashdot.








Book Review: Build Your Own Website: A Comic Guide to HTML, CSS, and WordPress

Slashdot - Wed, 17/12/2014 - 8:58pm
MassDosage writes "At the the risk of exposing my age I remember building my first website using a rudimentary Unix text editor (Joe) and carefully handcrafting the Hypertext Markup Language (HTML) while directly logged on to the web server it was being served from. Back then Cascading Style Sheets (CSS) weren't even a glint in the eyes of their creators. A lot has changed and there's now a world of fancy WYSIWYG web page editors to choose from as well as Content Management Systems that allow you to create websites without looking at the underlying code at all. While this is all very useful and allows less technical people to create websites I still feel that having at least some knowledge of how everything works under the hood is empowering — especially in situations where you want to go beyond the limits placed on you by a certain tool. This is where Build Your Own Website: A comic guide to HTML, CSS and Wordpress comes into the picture. Its aim is to enable people new to web development to learn the subject by teaching the fundamentals of HTML and CSS first and only then describing how to use a Content Management System (CMS) — in this case Wordpress. While Wordpress might not be everyone's kettle of fish it's a good choice as an example of a modern CMS that is easily accessible and very popular. The concepts presented are simple enough that it should be easy enough for a reader to apply them to a different CMS should they want to. Read below for The rest of MassDosage's review.

Read more of this story at Slashdot.








Book Review: Build Your Own Website: A Comic Guide to HTML, CSS, and WordPress

Slashdot - Wed, 17/12/2014 - 8:58pm
MassDosage writes "At the the risk of exposing my age I remember building my first website using a rudimentary Unix text editor (Joe) and carefully handcrafting the Hypertext Markup Language (HTML) while directly logged on to the web server it was being served from. Back then Cascading Style Sheets (CSS) weren't even a glint in the eyes of their creators. A lot has changed and there's now a world of fancy WYSIWYG web page editors to choose from as well as Content Management Systems that allow you to create websites without looking at the underlying code at all. While this is all very useful and allows less technical people to create websites I still feel that having at least some knowledge of how everything works under the hood is empowering — especially in situations where you want to go beyond the limits placed on you by a certain tool. This is where Build Your Own Website: A comic guide to HTML, CSS and Wordpress comes into the picture. Its aim is to enable people new to web development to learn the subject by teaching the fundamentals of HTML and CSS first and only then describing how to use a Content Management System (CMS) — in this case Wordpress. While Wordpress might not be everyone's kettle of fish it's a good choice as an example of a modern CMS that is easily accessible and very popular. The concepts presented are simple enough that it should be easy enough for a reader to apply them to a different CMS should they want to. Read below for The rest of MassDosage's review.

Read more of this story at Slashdot.








Book Review: Build Your Own Website: A Comic Guide to HTML, CSS, and WordPress

Slashdot - Wed, 17/12/2014 - 8:58pm
MassDosage writes "At the the risk of exposing my age I remember building my first website using a rudimentary Unix text editor (Joe) and carefully handcrafting the Hypertext Markup Language (HTML) while directly logged on to the web server it was being served from. Back then Cascading Style Sheets (CSS) weren't even a glint in the eyes of their creators. A lot has changed and there's now a world of fancy WYSIWYG web page editors to choose from as well as Content Management Systems that allow you to create websites without looking at the underlying code at all. While this is all very useful and allows less technical people to create websites I still feel that having at least some knowledge of how everything works under the hood is empowering — especially in situations where you want to go beyond the limits placed on you by a certain tool. This is where Build Your Own Website: A comic guide to HTML, CSS and Wordpress comes into the picture. Its aim is to enable people new to web development to learn the subject by teaching the fundamentals of HTML and CSS first and only then describing how to use a Content Management System (CMS) — in this case Wordpress. While Wordpress might not be everyone's kettle of fish it's a good choice as an example of a modern CMS that is easily accessible and very popular. The concepts presented are simple enough that it should be easy enough for a reader to apply them to a different CMS should they want to. Read below for The rest of MassDosage's review.

Read more of this story at Slashdot.








my product key have been stolen suddenly after 4 years of use

Windows Genuine Disadvantage [Vista] - Wed, 17/12/2014 - 8:35pm
Diagnostic Report (1.9.0027.0):
-----------------------------------------
Windows Validation Data-->
Validation Status: Genuine
Validation Code: 0
Cached Online Validation Code: N/A, hr = 0xc004f012
Windows Product Key: *****-*****-XK376-TJ7T4-76PKF
Windows Product Key Hash: 5QFdDDGgO8Pgmj93aZhOuZ69dH0=
Windows Product ID: 89578-022-5000027-71856
Windows Product ID Type: 5
Windows License Type: Retail
Windows OS version: 6.0.6002.2.00010300.2.0.003
ID: {3DA14DB2-2A95-4C44-9216-861830578883}(1)
Is Admin: Yes
TestCab: 0x0
LegitcheckControl ActiveX: N/A, hr = 0x80070002
Signed By: N/A, hr = 0x80070002
Product Name: Windows Vista (TM) Home Premium
Architecture: 0x00000000
Build lab: 6002.vistasp2_gdr.130707-1535
TTS Error: 
Validation Diagnostic: 
Resolution Status: N/A

Vista WgaER Data-->
ThreatID(s): N/A, hr = 0x80070002
Version: N/A, hr = 0x80070002

Windows XP Notifications Data-->
Cached Result: N/A, hr = 0x80070002
File Exists: No
Version: N/A, hr = 0x80070002
WgaTray.exe Signed By: N/A, hr = 0x80070002
WgaLogon.dll Signed By: N/A, hr = 0x80070002

OGA Notifications Data-->
Cached Result: N/A, hr = 0x80070002
Version: N/A, hr = 0x80070002
OGAExec.exe Signed By: N/A, hr = 0x80070002
OGAAddin.dll Signed By: N/A, hr = 0x80070002

OGA Data-->
Office Status: 109 N/A
OGA Version: N/A, 0x80070002
Signed By: N/A, hr = 0x80070002
Office Diagnostics: B4D0AA8B-604-645_77F760FE-153-80070002_7E90FEE8-175-80070002_B4D0AA8B-604-645_B4D0AA8B-604-645_025D1FF3-364-80041010_025D1FF3-229-80041010_025D1FF3-230-1_025D1FF3-517-80040154_025D1FF3-237-80040154_025D1FF3-238-2_025D1FF3-258-3

Browser Data-->
Proxy settings: N/A
User Agent: Mozilla/4.0 (compatible; MSIE 8.0; Win32)
Default Browser: C:\Program Files\Google\Chrome\Application\chrome.exe
Download signed ActiveX controls: Prompt
Download unsigned ActiveX controls: Disabled
Run ActiveX controls and plug-ins: Allowed
Initialize and script ActiveX controls not marked as safe: Disabled
Allow scripting of Internet Explorer Webbrowser control: Disabled
Active scripting: Allowed
Script ActiveX controls marked as safe for scripting: Allowed

File Scan Data-->

Other data-->
Office Details: <GenuineResults><MachineData><UGUID>{3DA14DB2-2A95-4C44-9216-861830578883}</UGUID><Version>1.9.0027.0</Version><OS>6.0.6002.2.00010300.2.0.003</OS><Architecture>x32</Architecture><PKey>*****-*****-*****-*****-76PKF</PKey><PID>89578-022-5000027-71856</PID><PIDType>5</PIDType><SID>S-1-5-21-4024307216-922108757-1920874934</SID><SYSTEM><Manufacturer>SAMSUNG ELECTRONICS CO., LTD.</Manufacturer><Model>R510/P510                  </Model></SYSTEM><BIOS><Manufacturer>Phoenix Technologies Ltd.</Manufacturer><Version>07LI.MP00.20080926.SCY</Version><SMBIOSVersion major="2" minor="5"/><Date>20080926000000.000000+000</Date></BIOS><HWID>54303507018400FA</HWID><UserLCID>1401</UserLCID><SystemLCID>0409</SystemLCID><TimeZone>Romance Standard Time(GMT+01:00)</TimeZone><iJoin>0</iJoin><SBID><stat>3</stat><msppid></msppid><name></name><model></model></SBID><OEM><OEMID>SECCSD</OEMID><OEMTableID>LH43STAR</OEMTableID></OEM><GANotification/></MachineData><Software><Office><Result>109</Result><Products/><Applications/></Office></Software></GenuineResults>  

Spsys.log Content: 0x80070002

Licensing Data-->
Software licensing service version: 6.0.6002.18005
Name: Windows(TM) Vista, HomePremium edition
Description: Windows Operating System - Vista, RETAIL channel
Activation ID: 9e042223-03bf-49ae-808f-ff37f128d40d
Application ID: 55c92734-d682-4d71-983e-d6ec3f16059f
Extended PID: 89578-00142-022-500002-00-5121-6002.0000-3022014
Installation ID: 000230133415864532277174875353734736756874157340680563
Processor Certificate URL: http://go.microsoft.com/fwlink/?LinkID=43473
Machine Certificate URL: http://go.microsoft.com/fwlink/?LinkID=43474
Use License URL: http://go.microsoft.com/fwlink/?LinkID=43476
Product Key Certificate URL: http://go.microsoft.com/fwlink/?LinkID=43475
Partial Product Key: 76PKF
License Status: Initial grace period
Time remaining: 15180 minute(s) (10 day(s))

Windows Activation Technologies-->
N/A

HWID Data-->
HWID Hash Current: MgAAAAEAAQABAAIAAQABAAAAAwABAAEAeqhcT55MGBS2W9KW6APy9HAyhCHCHaxWKoU=

OEM Activation 1.0 Data-->
N/A

OEM Activation 2.0 Data-->
BIOS valid for OA 2.0: yes
Windows marker version: 0x20000
OEMID and OEMTableID Consistent: yes
BIOS Information: 
  ACPI Table Name OEMID Value OEMTableID Value
  APIC PTLTD APIC  
  FACP INTEL CRESTLNE
  HPET INTEL CRESTLNE
  BOOT PTLTD $SBFTBL$
  MCFG INTEL CRESTLNE
  SLIC SECCSD LH43STAR
  SSDT BrtRef DD01BRT
  SSDT BrtRef DD01BRT
  SSDT BrtRef DD01BRT
  SSDT BrtRef DD01BRT

Research Highlights How AI Sees and How It Knows What It's Looking At

Slashdot - Wed, 17/12/2014 - 8:16pm
anguyen8 writes Deep neural networks (DNNs) trained with Deep Learning have recently produced mind-blowing results in a variety of pattern-recognition tasks, most notably speech recognition, language translation, and recognizing objects in images, where they now perform at near-human levels. But do they see the same way we do? Nope. Researchers recently found that it is easy to produce images that are completely unrecognizable to humans, but that DNNs classify with near-certainty as everyday objects. For example, DNNs look at TV static and declare with 99.99% confidence it is a school bus. An evolutionary algorithm produced the synthetic images by generating pictures and selecting for those that a DNN believed to be an object (i.e. "survival of the school-bus-iest"). The resulting computer-generated images look like modern, abstract art. The pictures also help reveal what DNNs learn to care about when recognizing objects (e.g. a school bus is alternating yellow and black lines, but does not need to have a windshield or wheels), shedding light into the inner workings of these DNN black boxes.

Read more of this story at Slashdot.








Research Highlights How AI Sees and How It Knows What It's Looking At

Slashdot - Wed, 17/12/2014 - 8:16pm
anguyen8 writes Deep neural networks (DNNs) trained with Deep Learning have recently produced mind-blowing results in a variety of pattern-recognition tasks, most notably speech recognition, language translation, and recognizing objects in images, where they now perform at near-human levels. But do they see the same way we do? Nope. Researchers recently found that it is easy to produce images that are completely unrecognizable to humans, but that DNNs classify with near-certainty as everyday objects. For example, DNNs look at TV static and declare with 99.99% confidence it is a school bus. An evolutionary algorithm produced the synthetic images by generating pictures and selecting for those that a DNN believed to be an object (i.e. "survival of the school-bus-iest"). The resulting computer-generated images look like modern, abstract art. The pictures also help reveal what DNNs learn to care about when recognizing objects (e.g. a school bus is alternating yellow and black lines, but does not need to have a windshield or wheels), shedding light into the inner workings of these DNN black boxes.

Read more of this story at Slashdot.








Research Highlights How AI Sees and How It Knows What It's Looking At

Slashdot - Wed, 17/12/2014 - 8:16pm
anguyen8 writes Deep neural networks (DNNs) trained with Deep Learning have recently produced mind-blowing results in a variety of pattern-recognition tasks, most notably speech recognition, language translation, and recognizing objects in images, where they now perform at near-human levels. But do they see the same way we do? Nope. Researchers recently found that it is easy to produce images that are completely unrecognizable to humans, but that DNNs classify with near-certainty as everyday objects. For example, DNNs look at TV static and declare with 99.99% confidence it is a school bus. An evolutionary algorithm produced the synthetic images by generating pictures and selecting for those that a DNN believed to be an object (i.e. "survival of the school-bus-iest"). The resulting computer-generated images look like modern, abstract art. The pictures also help reveal what DNNs learn to care about when recognizing objects (e.g. a school bus is alternating yellow and black lines, but does not need to have a windshield or wheels), shedding light into the inner workings of these DNN black boxes.

Read more of this story at Slashdot.








Research Highlights How AI Sees and How It Knows What It's Looking At

Slashdot - Wed, 17/12/2014 - 8:16pm
anguyen8 writes Deep neural networks (DNNs) trained with Deep Learning have recently produced mind-blowing results in a variety of pattern-recognition tasks, most notably speech recognition, language translation, and recognizing objects in images, where they now perform at near-human levels. But do they see the same way we do? Nope. Researchers recently found that it is easy to produce images that are completely unrecognizable to humans, but that DNNs classify with near-certainty as everyday objects. For example, DNNs look at TV static and declare with 99.99% confidence it is a school bus. An evolutionary algorithm produced the synthetic images by generating pictures and selecting for those that a DNN believed to be an object (i.e. "survival of the school-bus-iest"). The resulting computer-generated images look like modern, abstract art. The pictures also help reveal what DNNs learn to care about when recognizing objects (e.g. a school bus is alternating yellow and black lines, but does not need to have a windshield or wheels), shedding light into the inner workings of these DNN black boxes.

Read more of this story at Slashdot.








Research Highlights How AI Sees and How It Knows What It's Looking At

Slashdot - Wed, 17/12/2014 - 8:16pm
anguyen8 writes Deep neural networks (DNNs) trained with Deep Learning have recently produced mind-blowing results in a variety of pattern-recognition tasks, most notably speech recognition, language translation, and recognizing objects in images, where they now perform at near-human levels. But do they see the same way we do? Nope. Researchers recently found that it is easy to produce images that are completely unrecognizable to humans, but that DNNs classify with near-certainty as everyday objects. For example, DNNs look at TV static and declare with 99.99% confidence it is a school bus. An evolutionary algorithm produced the synthetic images by generating pictures and selecting for those that a DNN believed to be an object (i.e. "survival of the school-bus-iest"). The resulting computer-generated images look like modern, abstract art. The pictures also help reveal what DNNs learn to care about when recognizing objects (e.g. a school bus is alternating yellow and black lines, but does not need to have a windshield or wheels), shedding light into the inner workings of these DNN black boxes.

Read more of this story at Slashdot.








Research Highlights How AI Sees and How It Knows What It's Looking At

Slashdot - Wed, 17/12/2014 - 8:16pm
anguyen8 writes Deep neural networks (DNNs) trained with Deep Learning have recently produced mind-blowing results in a variety of pattern-recognition tasks, most notably speech recognition, language translation, and recognizing objects in images, where they now perform at near-human levels. But do they see the same way we do? Nope. Researchers recently found that it is easy to produce images that are completely unrecognizable to humans, but that DNNs classify with near-certainty as everyday objects. For example, DNNs look at TV static and declare with 99.99% confidence it is a school bus. An evolutionary algorithm produced the synthetic images by generating pictures and selecting for those that a DNN believed to be an object (i.e. "survival of the school-bus-iest"). The resulting computer-generated images look like modern, abstract art. The pictures also help reveal what DNNs learn to care about when recognizing objects (e.g. a school bus is alternating yellow and black lines, but does not need to have a windshield or wheels), shedding light into the inner workings of these DNN black boxes.

Read more of this story at Slashdot.








Research Highlights How AI Sees and How It Knows What It's Looking At

Slashdot - Wed, 17/12/2014 - 8:16pm
anguyen8 writes Deep neural networks (DNNs) trained with Deep Learning have recently produced mind-blowing results in a variety of pattern-recognition tasks, most notably speech recognition, language translation, and recognizing objects in images, where they now perform at near-human levels. But do they see the same way we do? Nope. Researchers recently found that it is easy to produce images that are completely unrecognizable to humans, but that DNNs classify with near-certainty as everyday objects. For example, DNNs look at TV static and declare with 99.99% confidence it is a school bus. An evolutionary algorithm produced the synthetic images by generating pictures and selecting for those that a DNN believed to be an object (i.e. "survival of the school-bus-iest"). The resulting computer-generated images look like modern, abstract art. The pictures also help reveal what DNNs learn to care about when recognizing objects (e.g. a school bus is alternating yellow and black lines, but does not need to have a windshield or wheels), shedding light into the inner workings of these DNN black boxes.

Read more of this story at Slashdot.








Research Highlights How AI Sees and How It Knows What It's Looking At

Slashdot - Wed, 17/12/2014 - 8:16pm
anguyen8 writes Deep neural networks (DNNs) trained with Deep Learning have recently produced mind-blowing results in a variety of pattern-recognition tasks, most notably speech recognition, language translation, and recognizing objects in images, where they now perform at near-human levels. But do they see the same way we do? Nope. Researchers recently found that it is easy to produce images that are completely unrecognizable to humans, but that DNNs classify with near-certainty as everyday objects. For example, DNNs look at TV static and declare with 99.99% confidence it is a school bus. An evolutionary algorithm produced the synthetic images by generating pictures and selecting for those that a DNN believed to be an object (i.e. "survival of the school-bus-iest"). The resulting computer-generated images look like modern, abstract art. The pictures also help reveal what DNNs learn to care about when recognizing objects (e.g. a school bus is alternating yellow and black lines, but does not need to have a windshield or wheels), shedding light into the inner workings of these DNN black boxes.

Read more of this story at Slashdot.








Research Highlights How AI Sees and How It Knows What It's Looking At

Slashdot - Wed, 17/12/2014 - 8:16pm
anguyen8 writes Deep neural networks (DNNs) trained with Deep Learning have recently produced mind-blowing results in a variety of pattern-recognition tasks, most notably speech recognition, language translation, and recognizing objects in images, where they now perform at near-human levels. But do they see the same way we do? Nope. Researchers recently found that it is easy to produce images that are completely unrecognizable to humans, but that DNNs classify with near-certainty as everyday objects. For example, DNNs look at TV static and declare with 99.99% confidence it is a school bus. An evolutionary algorithm produced the synthetic images by generating pictures and selecting for those that a DNN believed to be an object (i.e. "survival of the school-bus-iest"). The resulting computer-generated images look like modern, abstract art. The pictures also help reveal what DNNs learn to care about when recognizing objects (e.g. a school bus is alternating yellow and black lines, but does not need to have a windshield or wheels), shedding light into the inner workings of these DNN black boxes.

Read more of this story at Slashdot.








Research Highlights How AI Sees and How It Knows What It's Looking At

Slashdot - Wed, 17/12/2014 - 8:16pm
anguyen8 writes Deep neural networks (DNNs) trained with Deep Learning have recently produced mind-blowing results in a variety of pattern-recognition tasks, most notably speech recognition, language translation, and recognizing objects in images, where they now perform at near-human levels. But do they see the same way we do? Nope. Researchers recently found that it is easy to produce images that are completely unrecognizable to humans, but that DNNs classify with near-certainty as everyday objects. For example, DNNs look at TV static and declare with 99.99% confidence it is a school bus. An evolutionary algorithm produced the synthetic images by generating pictures and selecting for those that a DNN believed to be an object (i.e. "survival of the school-bus-iest"). The resulting computer-generated images look like modern, abstract art. The pictures also help reveal what DNNs learn to care about when recognizing objects (e.g. a school bus is alternating yellow and black lines, but does not need to have a windshield or wheels), shedding light into the inner workings of these DNN black boxes.

Read more of this story at Slashdot.








Research Highlights How AI Sees and How It Knows What It's Looking At

Slashdot - Wed, 17/12/2014 - 8:16pm
anguyen8 writes Deep neural networks (DNNs) trained with Deep Learning have recently produced mind-blowing results in a variety of pattern-recognition tasks, most notably speech recognition, language translation, and recognizing objects in images, where they now perform at near-human levels. But do they see the same way we do? Nope. Researchers recently found that it is easy to produce images that are completely unrecognizable to humans, but that DNNs classify with near-certainty as everyday objects. For example, DNNs look at TV static and declare with 99.99% confidence it is a school bus. An evolutionary algorithm produced the synthetic images by generating pictures and selecting for those that a DNN believed to be an object (i.e. "survival of the school-bus-iest"). The resulting computer-generated images look like modern, abstract art. The pictures also help reveal what DNNs learn to care about when recognizing objects (e.g. a school bus is alternating yellow and black lines, but does not need to have a windshield or wheels), shedding light into the inner workings of these DNN black boxes.

Read more of this story at Slashdot.








Research Highlights How AI Sees and How It Knows What It's Looking At

Slashdot - Wed, 17/12/2014 - 8:16pm
anguyen8 writes Deep neural networks (DNNs) trained with Deep Learning have recently produced mind-blowing results in a variety of pattern-recognition tasks, most notably speech recognition, language translation, and recognizing objects in images, where they now perform at near-human levels. But do they see the same way we do? Nope. Researchers recently found that it is easy to produce images that are completely unrecognizable to humans, but that DNNs classify with near-certainty as everyday objects. For example, DNNs look at TV static and declare with 99.99% confidence it is a school bus. An evolutionary algorithm produced the synthetic images by generating pictures and selecting for those that a DNN believed to be an object (i.e. "survival of the school-bus-iest"). The resulting computer-generated images look like modern, abstract art. The pictures also help reveal what DNNs learn to care about when recognizing objects (e.g. a school bus is alternating yellow and black lines, but does not need to have a windshield or wheels), shedding light into the inner workings of these DNN black boxes.

Read more of this story at Slashdot.








Research Highlights How AI Sees and How It Knows What It's Looking At

Slashdot - Wed, 17/12/2014 - 8:16pm
anguyen8 writes Deep neural networks (DNNs) trained with Deep Learning have recently produced mind-blowing results in a variety of pattern-recognition tasks, most notably speech recognition, language translation, and recognizing objects in images, where they now perform at near-human levels. But do they see the same way we do? Nope. Researchers recently found that it is easy to produce images that are completely unrecognizable to humans, but that DNNs classify with near-certainty as everyday objects. For example, DNNs look at TV static and declare with 99.99% confidence it is a school bus. An evolutionary algorithm produced the synthetic images by generating pictures and selecting for those that a DNN believed to be an object (i.e. "survival of the school-bus-iest"). The resulting computer-generated images look like modern, abstract art. The pictures also help reveal what DNNs learn to care about when recognizing objects (e.g. a school bus is alternating yellow and black lines, but does not need to have a windshield or wheels), shedding light into the inner workings of these DNN black boxes.

Read more of this story at Slashdot.








Research Highlights How AI Sees and How It Knows What It's Looking At

Slashdot - Wed, 17/12/2014 - 8:16pm
anguyen8 writes Deep neural networks (DNNs) trained with Deep Learning have recently produced mind-blowing results in a variety of pattern-recognition tasks, most notably speech recognition, language translation, and recognizing objects in images, where they now perform at near-human levels. But do they see the same way we do? Nope. Researchers recently found that it is easy to produce images that are completely unrecognizable to humans, but that DNNs classify with near-certainty as everyday objects. For example, DNNs look at TV static and declare with 99.99% confidence it is a school bus. An evolutionary algorithm produced the synthetic images by generating pictures and selecting for those that a DNN believed to be an object (i.e. "survival of the school-bus-iest"). The resulting computer-generated images look like modern, abstract art. The pictures also help reveal what DNNs learn to care about when recognizing objects (e.g. a school bus is alternating yellow and black lines, but does not need to have a windshield or wheels), shedding light into the inner workings of these DNN black boxes.

Read more of this story at Slashdot.








Research Highlights How AI Sees and How It Knows What It's Looking At

Slashdot - Wed, 17/12/2014 - 8:16pm
anguyen8 writes Deep neural networks (DNNs) trained with Deep Learning have recently produced mind-blowing results in a variety of pattern-recognition tasks, most notably speech recognition, language translation, and recognizing objects in images, where they now perform at near-human levels. But do they see the same way we do? Nope. Researchers recently found that it is easy to produce images that are completely unrecognizable to humans, but that DNNs classify with near-certainty as everyday objects. For example, DNNs look at TV static and declare with 99.99% confidence it is a school bus. An evolutionary algorithm produced the synthetic images by generating pictures and selecting for those that a DNN believed to be an object (i.e. "survival of the school-bus-iest"). The resulting computer-generated images look like modern, abstract art. The pictures also help reveal what DNNs learn to care about when recognizing objects (e.g. a school bus is alternating yellow and black lines, but does not need to have a windshield or wheels), shedding light into the inner workings of these DNN black boxes.

Read more of this story at Slashdot.








Syndicate content