TED talk link -- http://www.ted.com/talks/luis_von_ahn_massive_scale_online_collaboration
In this TED talk, it introduces how computer scientists leverage the time of millions of people for human good. Prof. Luis von Ahn from CMU designed a system called CAPTCHA, which ask people to recognize some distorted words when they register an account online. After that, he realized 10 seconds spend on recognizing characters could be applied to digitize the books. That's why reCAPTCHA comes. It offers two words. Computer has result of one word, which use as the same function of CAPTCHA. And another one is the book that need to be digize. When people type answers for those words, they help to recognize one of billions of words for digitizing books.
Suddenly, I realized what happens behind the scenes when I was asked to input a disordered sequence of characters in account registration. It seems like everyone in the internet makes contributions to the digitize human knowledge. 10 seconds for a person may be trivial, but when it comes to millions of people, it could do some great things. Luis' idea fills up the deficiency of computer algorithm and split the complex task into pieces. The result is amazing.
Similarly, Duolingo has already been a popular word for a long time. But now I am aware of its huge social value. It definitely would be sustainable business model. People only need spend time on it to obtain high quality of learning resource. It means a lot to the people in developing world, where people have time but not money.
This is a mind-open lecture. I believe with the same idea, in the future, there are more and more application which use million people's small effort to make impossible things possible.
Take Home Message
Thursday, November 20, 2014
Sunday, November 16, 2014
Powering off your phone is not a sustainable strategy to fight against information overloading
Powering off your phone is not a sustainable strategy to fight against information overloading
Gradually, the only possible way to motivate me to insist on something is that I can create a clear message from my heart. It's a clear understanding towards what you really want. I don't need too much useless information and instead, I need time to do more thinking on the existing information I have. Someone would say if you don't catch up the step of the era, you would be left behind.
However, after I spent enough time on reading, I found the greatest minds are always in such a few books. And most of the newest information in the world is just the poor replication of them. What's more, in order to fully understand these thoughts, there must be considerable efforts. I really don't want to be disturbed by the new but poor information.
Sunday, September 28, 2014
Take home messages after talking with a libarian
How to find a paper related to MOOC?
It's better Go to the education database in UVA.
The topic list from UVA library -- http://guides.lib.virginia.edu/subjectguides
Topic Computer science --http://guides.lib.virginia.edu/content.php?pid=16870&sid=131270
The trick to use the keywords
1. After you search a roughly word, you may need find a keyword in su
2. we could use "Direct Export to RefWorks" to manage our papers
3. Use "Limiters - Scholarly (Peer Reviewed) Journals"
Add filter --"peer reviewed", and it will allow us to get papers with higher quality.
4. Use the subject message in one paper as the searching keywords
5. Search the keyword in subject message, title or abstract.
6. You could also search the paper by author's name and then from the result, you could see the research focus and find their proficient area.
7. If you want to search about the impactor rate for a certain journal or conference. You need find it here:
It's better Go to the education database in UVA.
The topic list from UVA library -- http://guides.lib.virginia.edu/subjectguides
Topic Computer science --http://guides.lib.virginia.edu/content.php?pid=16870&sid=131270
The trick to use the keywords
1. After you search a roughly word, you may need find a keyword in su
2. we could use "Direct Export to RefWorks" to manage our papers
3. Use "Limiters - Scholarly (Peer Reviewed) Journals"
Add filter --"peer reviewed", and it will allow us to get papers with higher quality.
4. Use the subject message in one paper as the searching keywords
5. Search the keyword in subject message, title or abstract.
6. You could also search the paper by author's name and then from the result, you could see the research focus and find their proficient area.
7. If you want to search about the impactor rate for a certain journal or conference. You need find it here:
8. after your searching, you may find there is no available pdf version, you may consider about use other ways to search. For example, google, or ISBN number and so on.
Friday, September 19, 2014
Exchange two variables without usage of extra memory
Usually, when we want to exchange two variables, we would think of adding a temp variable. However, there is a way to avoiding the use of extra memory.
Example:
x =a, y = b
step 1:
x := x +y
x = a+b, y =b
step 2:
y := x -y =(a+b) - b = a
x = a+b, y=a
step 1:
x := x +y
x = a+b, y =b
step 2:
y := x -y =(a+b) - b = a
x = a+b, y=a
step 3:
x := x - y = (a+b) - a = b
x = b, y=a
x := x - y = (a+b) - a = b
x = b, y=a
The algorithm to initializing an array
Quick notes from website--http://eli.thegreenplace.net/2008/08/23/initializing-an-array-in-constant-time/
The problem
We want to use a very large array for some computations. When created, all the elements of this array have to be initialized to some value. We'll only use a few values from the array1, so we don't want the runtime of the algorithm to be dominated by the initialization time.
In other words, we want to create and access an initialized array in constant time2.
How can this be done ? (Hint: we may use extra memory)
Solution
For a n-element array (
vec), we'll use two additional n-element arrays and a pointer3. The arrays are called from andto and the pointer is called top. init_val is the value we want to initialize the array with.- When an element
ifromvecis accessed, we'll know whether it has been initialized iffrom[i] < top and to[from[i]] = i. If this condition is satisfied, we returnvec[i]. Otherwise, we returninit_val. - Initially,
topis 0 - The array element
iis first accessed by the code:
from[i] = top to[top] = i vec[i] = init_val top++
Understanding the algorithm
Let's assume we want an array with size 10. This is the initial state of the array and helpers (
X denotes a memory location with arbitrary contents):vec: X X X X X X X X X X from: X X X X X X X X X X to: X X X X X X X X X X top: 0
Since
from and to contain unsigned values, the condition from[i] < top can never be true while top = 0. So all read accesses at this point will result in init_val.
Now suppose we write into the vector. Let's say
vec[4] <- 3. According to the algorithm, the next state will be:vec: X X X X 3 X X X X X from: X X X X 0 X X X X X to: 4 X X X X X X X X X top: 1
Now accesses to
i = 4 will satisfy the conditions from[4] = 0 < 1 and to[from[4]] = 4, so the value of vec[4] can be returned. Accesses to other indices will fail, because the only i that will pass the from[i] < top is an i for whichfrom[i] = 0, but only i = 4 will also fulfill the second condition to[from[i]] = i.
Let's work through another step. Let's say we write
vec[9] <- 6. The next state will be:vec: X X X X 3 X X X X 6 from: X X X X 0 X X X X 1 to: 4 9 X X X X X X X X top: 2
Again, the accesses to initialized indices,
i = 4, 9 will satisfy the conditions. Accesses to other indices will fail. Suppose for example an access to i = 7. Is from[7] < 2 ? Let's assume that by chance it is. It must be either 0 or 1 then. But for both, to points to the only really initialized fields, which are either 4 or 9, so the second condition will not be satisfied.
I hope it's now clear how
to, from and top interact to ensure that only initialized values will be actually returned. They kind-of close a loop - i points to from, from points to to and to points back to i. top makes sure that the location into was already initialized by the algorithm. Since the relevant locations in from and to are always in a controlled state, this algorithm can never be fooled by random values that were just lying around in the memory.Tuesday, September 16, 2014
Video or Text?
If you want to introduce about how female do in Computer Science. Video may not be a best solution.
For example:
http://www.dotdiva.org/
http://www.dotdiva.org/profiles.html
Text is faster to be understood by some people. Offering both text and video would be better!
For example:
http://www.dotdiva.org/
http://www.dotdiva.org/profiles.html
Text is faster to be understood by some people. Offering both text and video would be better!
Subscribe to:
Posts (Atom)



