A quick Google search of the NEWS for the word "Deep Learning" will reveal the strides Microsoft has made in the area of - "Crawling the Internet and Making decisions based on Probability estimates" so that we can use a cell photo to take a photo which will somehow say this is most likely a Welsh corgis - clearly discriminating Pembroke Welsh corgi from the Cardigan Welsh corgi ( LINK HERE)
Now lets get some facts out of the way. We all know that Facebook has been leading with this sort of technology ( AI or ML ( Deep Learning)) ( the details are not well known, but we know that they have one of the largest labeled data set for training with photos , mostly of humans).
Seconds, we know that Google did some announcements about their success in the training of a 16000 computer to recognize CATS, using network of network of information, through videos.
Now Microsoft is blowing their horn on this topic, revealing that they are doing image recognition through the deep learning process, with Project ADAM, I hope this does not refer to the name of the first man in the Bible.
Now According to Warburton (2003) " Deep learning is a key strategy by which students extract meaning and understanding from course materials and experiences. "
As an educator I may be a bit confused with the term "deep learning" as applied to Artificial Intelligence verses the student.
There is little to be confused about. The principles are the same and in fact, the term is being used because it is the way we learn, by connecting experience, information and meaning from one situation to another.
Now the application of these concepts from ( Education and Learning theories) are being successfully done in the area of Artificial Intelligence ( AI).
However, this is where we need dig deeper. First of all, the current methods may be limited, the transfer of " knowledge" from one subject domain to another is not achieved - unlike a real human (off course we are in the early stage of this break through).
The current application of these technologies are in recognition systems - photos, audio,etc. Area where users don't need to worry about a "poor result" and where information on the Internet can be easily accesses with predefined labels.
Which leaves me to wonder. If the training set is always the mixed of labeled photos/images, how can we do deep learning with text, information that is actually the most available on the internet?
In Conclusion, I believe lots of the focus is on the vision of the computer..... while the brain is being designed as we play along..... However, it will be interesting to see which of these companies ( including Apple, Yahoo and the many Universities ) will actually develop a full brain.....
Reference list and links:
 |
| Image from Google News Results - Dated 17/07/2014 6:45 AM ( Australian Time) |
Now lets get some facts out of the way. We all know that Facebook has been leading with this sort of technology ( AI or ML ( Deep Learning)) ( the details are not well known, but we know that they have one of the largest labeled data set for training with photos , mostly of humans).
Seconds, we know that Google did some announcements about their success in the training of a 16000 computer to recognize CATS, using network of network of information, through videos.
Now Microsoft is blowing their horn on this topic, revealing that they are doing image recognition through the deep learning process, with Project ADAM, I hope this does not refer to the name of the first man in the Bible.
Now According to Warburton (2003) " Deep learning is a key strategy by which students extract meaning and understanding from course materials and experiences. "
As an educator I may be a bit confused with the term "deep learning" as applied to Artificial Intelligence verses the student.
There is little to be confused about. The principles are the same and in fact, the term is being used because it is the way we learn, by connecting experience, information and meaning from one situation to another.
Now the application of these concepts from ( Education and Learning theories) are being successfully done in the area of Artificial Intelligence ( AI).
However, this is where we need dig deeper. First of all, the current methods may be limited, the transfer of " knowledge" from one subject domain to another is not achieved - unlike a real human (off course we are in the early stage of this break through).
The current application of these technologies are in recognition systems - photos, audio,etc. Area where users don't need to worry about a "poor result" and where information on the Internet can be easily accesses with predefined labels.
Which leaves me to wonder. If the training set is always the mixed of labeled photos/images, how can we do deep learning with text, information that is actually the most available on the internet?
In Conclusion, I believe lots of the focus is on the vision of the computer..... while the brain is being designed as we play along..... However, it will be interesting to see which of these companies ( including Apple, Yahoo and the many Universities ) will actually develop a full brain.....
Reference list and links:
Warburton, Kevin. "Deep learning and education for sustainability." International Journal of Sustainability in Higher Education 4.1 (2003): 44-56.
Phys Blog announcement of Microsoft Deep Learning project.
http://phys.org/news/2014-07-welsh-corgis-vision-power-deep.html
IMI CHITTERMAN
Student, Author and wirting of all thing Grenadian
Contact Imi on mychitterman@gmail.com
Comments
Post a Comment