Definition of a function and a transform

1. Definition of a function: a function is a mapping of one set of numbers to another set of numbers. The first set is called the domain and the other set is called the co-domain. A function has to satisfy two properties.

The first one is that every element in the domain must respond to an element in the co-domain.

The second is that no two elements in the co-domain respond to the same element in the domain.

2. Definition of a transform: a transform is the mapping of a set of functions to another set of functions.

Object oriented programming 1 in java: objects and methods

I/ Declare variables and initialize string objects

1. String s1;// String is a class
2. String s1 = new String(); // create an empty string s1
3. String s1 = "Whatever"; // every string is an object
4. String s2 = s1; // s2 references to a box (object) that has a content of "Whatever"
5. String s2 = new String(s1);

II/ Methods

1. String s3= s1.Uppercase() // s3 references to a box that has a content of "WHATEVER"

How to buy a mutimedia projector?

1. Brightness: 2000 ANSI lumens or more
2. Contrast: 2000:1 or more
3. Aspect Ratio - 4:3 ; 16:9
4. Bulb time: 6000 hours or more
5. XGA or at least SVGA resolution (800 x 600)
6. Display color support: 24-bit (16.7 million colors)
7. Zoom: digital and optical
8. Project TV, DVD, PC and laptop
9. Price: less than $A1000
10. Supports HD signals including 720p and 1080i
11. Minimum image size (diagonal): 1m and maximum image size: 7m
12. Minimum screen distance: 2m and maximum screen distance: 6m
13. Weight: 5kg or less
14. Technology: DLP (digital light processing), 3LCD
15. Hardware platform: PC and Mac
16. Audible noise: 22dB or less
17. Projector types: business or home use
18. Brands: viewsonic, benq, acer, nec, sharp, panasonic, epson, sanyo, hp, jvc, toshiba, optoma
19. Possible products: TOSHIBA TDP SP1 DLP Projector ($A600) for business, BENQ MP612 ($800) for home use

Data visualisation techniques in data mining

1. Parallel coordinates
2. Scatter plots
3. Trent figures
4. Facial features expression

Where are they applied?

1. To speed up product design process
2. To analyze conformation of chemical structure
3. To measure quality of life (e.g. to visualise how different factors affect the degree of happiness)

Teamwork

1. Teamwork size should be about 3,4 or 5 people (i.e. too many people make it hard to coordinate)
2. Teamwork should be diversity in ages (i.e. to generate several perspectives)
3. Teamwork should be diversity in professions (i.e. to compensate different skills set)

Philosophy of business entrepreneurs

New product development should focus on the following things.

1. To improve quality of life.
2. To invent a new way of doing things.
3. Prevent something good from ending.
4. Have a passion for what we do.

K-nearest neighbour classfier, k-fold cross validation, leave-one-out

1. K-nearest neighbour classfier
Base on k training instances to classify a new instance into the majority class. The distance between the new instance and each of k training instances is often computed by Euclidean distance.

2. K-fold cross validation
The idea of k-fold cross validation is to try to estimate the true predictive accuracy of a classifier. It is often carried out for a small training dataset. Instead of using a separate small test dataset, we can divide the training dataset into k portions and then use 1 portion for testing and k-1 for training, and then repeat for different portions. Doing that, a more accurate predictive accuracy of a classifier can be achieved.

3. Leave-one-out cross validation
Instead of dividing the training dataset into k different portions, only one instance is kept for testing and the rest is for training.

Philosophy of fear, change and risk taking

1. Everyone has some kind of fear.
2. Almost everyone fear of change.
3. To overcome this we need to show them the vision why the change is worth taking risks.

Philosophy of fear, change and risk taking

1. Everyone has some kind of fear.
2. Almost everyone fear of change.
3. To overcome this we need to show them the vision why the change is worth taking risks.

Zippo lighter

Lessons from Zippo lighter

1. Tried to cooperate with a Swiss watch company but failed.
2. Introduced new products such as belts, pens but failed.
3. Lifetime warranty is an advantage.
4. New marketing strategy to young people in live music concerts and succeeded.
5. Targeted to young women with different designs and succeeded.

Zippo lighter

Lessons from Zippo lighter

1. Tried to cooperate with a Swiss watch company but failed.
2. Introduced new products such as belts, pens but failed.
3. Lifetime warranty is an advantage.
4. New marketing strategy to young people in live music concerts and succeeded.
5. Targeted to young women with different designs and succeeded.

Event organiser

Event organiser needs to organise the following

1. Place
2. MC
3. Catering
4. Advertising banners
5. Music band and singers
6. Designing and sending out invitation cards, brochures
7. Media announcement
8. Time scheduling
9. Security arrangement

Event organiser

Event organiser needs to organise the following

1. Place
2. MC
3. Catering
4. Advertising banners
5. Music band and singers
6. Designing and sending out invitation cards, brochures
7. Media announcement
8. Time scheduling
9. Security arrangement

Castling rules

1. Castle King side
2. Castle Queen side

The following rules apply if you want to castle

1. The King and the Rook must have never been moved before the King is allowed to be castling.
2. No block between the King and the Rook.
3. The King is not in check.
4. The spaces between the King and the Rook are not under attack.

When castling, the King must move first and then the Rook.

Castling rules

1. Castle King side
2. Castle Queen side

The following rules apply if you want to castle

1. The King and the Rook must have never been moved before the King is allowed to be castling.
2. No block between the King and the Rook.
3. The King is not in check.
4. The spaces between the King and the Rook are not under attack.

When castling, the King must move first and then the Rook.

How to buy a digital photo frame

To buy a digital photo frame please check the following features whether it has

1. MP3/MP4 player
2. Memory card slot and maximum memory size support (e.g. SD, USB stick)
3. Screen size (e.g. 7 inches or others)
4. Brightness, color saturation, contrast level
5. Warranty period
6. Batteries and/or ac adapter
7. Brands (e.g. Kodak, Laser)

How to buy a digital photo frame

To buy a digital photo frame please check the following features whether it has

1. MP3/MP4 player
2. Memory card slot and maximum memory size support (e.g. SD, USB stick)
3. Screen size (e.g. 7 inches or others)
4. Brightness, color saturation, contrast level
5. Warranty period
6. Batteries and/or ac adapter
7. Brands (e.g. Kodak, Laser)

How to buy a GPS device

To buy a GPS device please check the following features whether it has

1. Blue tooth
2. Touchscreen
3. Screen size 4.5 inch or greater
4. MP3/MP4 player
5. Memory card slot (e.g. SD)
6. Australian latest map
7. Speed camera alert
8. Red light camera alert
9. Hot spots (e.g. restaurants, supermarkets, petrol stations, cinemas, shopping centers)
10. Brands

How to buy a GPS device

To buy a GPS device please check the following features whether it has

1. Blue tooth
2. Touchscreen
3. Screen size 4.5 inch or greater
4. MP3/MP4 player
5. Memory card slot (e.g. SD)
6. Australian latest map
7. Speed camera alert
8. Red light camera alert
9. Hot spots (e.g. restaurants, supermarkets, petrol stations, cinemas, shopping centers)
10. Brands

How to buy an old car

To buy an old car please check the following features whether it has

1. Brands
2. Less than 10 years
3. Less than 100K km
4. Less than 2.4 lit engine
5. Automatic transmission
6. RWC
7. Rego included
8. Price less than $AUS 10,000
9. Service history
10. Airbags
11. Power steering
12. Central locker
13. Power mirrors
14. Cruise control
15. ABS system
16. Hifi system
17. Check title clearance with RACV

How to buy an old car

To buy an old car please check the following features whether it has

1. Brands
2. Less than 10 years
3. Less than 100K km
4. Less than 2.4 lit engine
5. Automatic transmission
6. RWC
7. Rego included
8. Price less than $AUS 10,000
9. Service history
10. Airbags
11. Power steering
12. Central locker
13. Power mirrors
14. Cruise control
15. ABS system
16. Hifi system
17. Check title clearance with RACV

Causation vs. correlation

We are interested in causation but usually we mistakes it with correlation.

1. For example, wearing red clothes does not mean lucky. There might be several instances that they are occurred together but wearing red cloth does not cause you good luck or bad luck.

2. In order to test if one thing causes another we can apply the control group in experiments like the way they test new drugs.

-For example, a new drug for treating depression is being developed, however we need to test if it causes the patient no more depression? We do this by having two groups of "depressed" patients. One is given the new drug and the other is given a placebo (i.e. the drug is just a candy). By analysing the results from these two groups we can decide if the new drug is the cause for treating the patient or not.

Causation vs. correlation

We are interested in causation but usually we mistakes it with correlation.

1. For example, wearing red clothes does not mean lucky. There might be several instances that they are occurred together but wearing red cloth does not cause you good luck or bad luck.

2. In order to test if one thing causes another we can apply the control group in experiments like the way they test new drugs.

-For example, a new drug for treating depression is being developed, however we need to test if it causes the patient no more depression? We do this by having two groups of "depressed" patients. One is given the new drug and the other is given a placebo (i.e. the drug is just a candy). By analysing the results from these two groups we can decide if the new drug is the cause for treating the patient or not.

Intellectual property vs. copyright

1. Intellectual property law is used to protect the author's creations and creativities. The author deserves financial benefits as well as moral benefits for their creations such as performances, books, patents, trademarks, copyrights, lecture notes, songs, poem.

2. Copyright is a document that allows one to publish and sell literary, musical or artistic work.

Intellectual property vs. copyright

1. Intellectual property law is used to protect the author's creations and creativities. The author deserves financial benefits as well as moral benefits for their creations such as performances, books, patents, trademarks, copyrights, lecture notes, songs, poem.

2. Copyright is a document that allows one to publish and sell literary, musical or artistic work.

News 02/10/2008

Headlines in Vietnam 02/10/2008

1. The US embassy in Hanoi announced three projects totaling 1 million USD to help the disabled in the central city of Danang to access healthcare services and develop a range of life skills. This is a part of the US Government's commitments to a 3 millionn USD budget to help the disabled suffering of the effects of dioxins. This event is following the annual conference about Agent Orange about two-three weeks ago.

2. The hospitals in Hanoi and Hochiminh City are overloaded with parents who have brought their infants for kydney check-up because they fear that their children had drunk milk containing melamine.
Domestic suppliers of milk-based products are rushing to laboratories for melamine tests on their products. The Health Ministry said around 500 milk samples were sent to four main testing centres in Hanoi and Hochiminh City for melamine detection. The result would be known within two weeks. Also, the ministry published a list of 87 dairy products without melamine on Monday.

3. Malaysia heads a list of 40 contries and territories who have invested in Vietnam over the last nine months with the project to contruct a steel complex cost 9.8 billion USD in Ninh Thuan. This is the biggest foreign invesment project in Vietnam so far.

4. Deputy Prime Minister Nguyen Thien Nhan approved the organisation of the beauty contest Miss World 2010 in Nha Trang. This is the second international beauty contest that is organised in Vietnam, following the Miss Universe this year.

The Word Today

1. The main story around the finacial crisis in US. The Senate will vote on the 700 billion US rescue plan this afternoon. The AM program of ABC this morning had an interview with David Hale, Global Economic Advisor for the Commonwealth Bank, about an overview this crisis, what are the implications of this global economic meltdown for our resource-driven economy.

2. US Christopher Hill is visiting North Korea to try to break nuclear program deadlock.

3. Australian new chief scientist has pointed the issue of climate change, water and sustainable energy as the key research that Australia is facing.

4. Once every half a century, a species of bamboo bursts into flower in some Asian countries like Cambodia, Indonesia and Burma attract huge number of rats. They are devastating crops and destroying livelihoods.

Automatic music emotions classification

There are 8 categories of music emotions.

1. Sad
2. Happy
3. Interesting

and there are 5 mores.

The following features are used to classify music emotions

1. Tempo
2. Pitches

Automatic music emotions classification

There are 8 categories of music emotions.

1. Sad
2. Happy
3. Interesting

and there are 5 mores.

The following features are used to classify music emotions

1. Tempo
2. Pitches

Telecommunications evolution

1. Network transformation (e.g. time-based switches to packet-based switches).
2. Service providers transformation (e.g. voice and limited data service on public switched telephone network (PSTN) vs. voice and broadband-based voice and data on fixed and wireless networks).
3. Service transformation (e.g. voice over copper to voice over IP networks, TVs over IP networks, tele-video conferences).

Telecommunications evolution

1. Network transformation (e.g. time-based switches to packet-based switches).
2. Service providers transformation (e.g. voice and limited data service on public switched telephone network (PSTN) vs. voice and broadband-based voice and data on fixed and wireless networks).
3. Service transformation (e.g. voice over copper to voice over IP networks, TVs over IP networks, tele-video conferences).

Data representation

To represent a source of data in a compact form
1. Using the affine transform
2. Using the fractal transform
3. Using the Fourier transform
4. Using the wavelet transform

Data representation

To represent a source of data in a compact form
1. Using the affine transform
2. Using the fractal transform
3. Using the Fourier transform
4. Using the wavelet transform

Transform coding

1. Divide the sequence {x_n} into N blocks. Each block is mapped into a transformed sequence using a reversible mapping.
2. Quantizing the transform sequence. The quantization strategy depends on three main factors: the desired average bit rate, the statistics of the elements of transformed sequence and the effect of distortion in the transformed sequence on the reconstructed sequence.
3. Encode the quantized values using some binary encoding technique.

Transform coding

1. Divide the sequence {x_n} into N blocks. Each block is mapped into a transformed sequence using a reversible mapping.
2. Quantizing the transform sequence. The quantization strategy depends on three main factors: the desired average bit rate, the statistics of the elements of transformed sequence and the effect of distortion in the transformed sequence on the reconstructed sequence.
3. Encode the quantized values using some binary encoding technique.

Vector quantization

I. How is VQ designed?

An image is divided into different blocks (e.g. 4x4 pixels). We use a codebook of length 16, 32, 64, 256. Sending the codebook, the index of the codebook to the receiver for decompression. Initialization for the codebook consists of 2 different strategies

1. Choose one vector as the averaged vector of the training set. Then split the training set into two sets, four sets and eight sets according to the requirements of the length of the codebook (i.e. 16, 32, 64)
2. Combine two vectors into one if they contribute a smallest distortion, continue until we have the required codebook.

II. What is VQ used for?

-Speech coding
-Image compression in low bit rates

Vector quantization

I. How is VQ designed?

An image is divided into different blocks (e.g. 4x4 pixels). We use a codebook of length 16, 32, 64, 256. Sending the codebook, the index of the codebook to the receiver for decompression. Initialization for the codebook consists of 2 different strategies

1. Choose one vector as the averaged vector of the training set. Then split the training set into two sets, four sets and eight sets according to the requirements of the length of the codebook (i.e. 16, 32, 64)
2. Combine two vectors into one if they contribute a smallest distortion, continue until we have the required codebook.

II. What is VQ used for?

-Speech coding
-Image compression in low bit rates

New trends in designing mobile phones

1. Each mobile phone can become a wifi hot-spot which can be used for several users to connect to the Internet.
2. Locations-based services (e.g. movies, restaurants, petrol stations, games centers, shopping centers, etc.).
3. More features are added (e.g. HD video recorders, high resolution cameras, higher optical zooms, etc.).
4. Longer battery life.
5. Prettier appearance (e.g. different shapes and forms).

New trends in designing mobile phones

1. Each mobile phone can become a wifi hot-spot which can be used for several users to connect to the Internet.
2. Locations-based services (e.g. movies, restaurants, petrol stations, games centers, shopping centers, etc.).
3. More features are added (e.g. HD video recorders, high resolution cameras, higher optical zooms, etc.).
4. Longer battery life.
5. Prettier appearance (e.g. different shapes and forms).

Mathematical spaces and other concepts

1. Metric spaces (e.g. scalars, vectors, sequences, functions)
2. Norm spaces
3. Vector spaces
4. Inner product spaces
5. Functions -> sequence of functions
6. Number -> sequence of numbers

Mathematical spaces and other concepts

1. Metric spaces (e.g. scalars, vectors, sequences, functions)
2. Norm spaces
3. Vector spaces
4. Inner product spaces
5. Functions -> sequence of functions
6. Number -> sequence of numbers

Mathematical spaces and other concepts

1. Metric spaces (e.g. scalars, vectors, sequences, functions)
2. Norm spaces
3. Vector spaces
4. Inner product spaces
5. Functions -> sequence of functions
6. Number -> sequence of numbers

Naive Bayes classifier

1. Continuous numerical attributes need to use Normal distribution to compute P(income=12000|Yes), etc. The set of values of the continuous numerical attribute is used to find out the mean and variance.

Naive Bayes classifier

1. Continuous numerical attributes need to use Normal distribution to compute P(income=12000|Yes), etc. The set of values of the continuous numerical attribute is used to find out the mean and variance.

K-means and k-modes

1. When compare two values of a nominal attribute if they are equal the result is zero otherwise the result will be 1/n (n is the number of values of this nominal attribute).

K-means and k-modes

1. When compare two values of a nominal attribute if they are equal the result is zero otherwise the result will be 1/n (n is the number of values of this nominal attribute).

Upcomming new technologies and other concepts

1. USB3 devices (10x faster than USB2: about 4.8Gbps)
2. Cloud computing
3. Machine learning involves how to write a program that can learn. It learns from examples and feedback. Machine learning techniques include neural networks. It usually deals with a small set of data. The data must be error-free and must be appropriate for the learning task. An example application of machine learning is the computer chess program.
4. Data mining concerns about how to extract knowledge from existing data. The output of data mining queries is the rules, classification or clusters.

Upcomming new technologies and other concepts

1. USB3 devices (10x faster than USB2: about 4.8Gbps)
2. Cloud computing
3. Machine learning involves how to write a program that can learn. It learns from examples and feedback. Machine learning techniques include neural networks. It usually deals with a small set of data. The data must be error-free and must be appropriate for the learning task. An example application of machine learning is the computer chess program.
4. Data mining concerns about how to extract knowledge from existing data. The output of data mining queries is the rules, classification or clusters.

Trends in designing new laptops and how to buy them

1. Better battery time
2. Prettier appearance (e.g. color, material, etc.)
3. Lighter
4. Faster

Buying a new laptop needs to consider the followings

1. Battery time
2. Weight
3. Memory (e.g. max 4GB for current 32 bits OS)
4. CPU (e.g. duo core, core 2 duo, quad core, atom, etc)
5. OS (e.g. XP, Vista, Linux, etc.)
6. Brand (e.g. Toshiba, IBM, Sony, Acer, Fujitsu, Dell, HP, etc.)

Trends in designing new laptops and how to buy them

1. Better battery time
2. Prettier appearance (e.g. color, material, etc.)
3. Lighter
4. Faster

Buying a new laptop needs to consider the followings

1. Battery time
2. Weight
3. Memory (e.g. max 4GB for current 32 bits OS)
4. CPU (e.g. duo core, core 2 duo, quad core, atom, etc)
5. OS (e.g. XP, Vista, Linux, etc.)
6. Brand (e.g. Toshiba, IBM, Sony, Acer, Fujitsu, Dell, HP, etc.)

How to design an optimal code

1. Kraft's inequality theorem tells us that if the Kraft's condition is satisfied, there exists a prefix code C. Also, if the code C is the prefix code, the Kraft's condition is satisfied.

2. The average length of the code is always greater than the entropy of the source S. Therefore, the better code is when L(C) is closer to H(S).

H(S) <= L(C)

where the source S has the probability distribution as P(S)={p1, p2, p3,..,pn}
L(C)=Sum of {l1 x p1 + l2 x p2 + l3 x p3 + ...+ ln x pn}

How to design an optimal code

1. Kraft's inequality theorem tells us that if the Kraft's condition is satisfied, there exists a prefix code C. Also, if the code C is the prefix code, the Kraft's condition is satisfied.

2. The average length of the code is always greater than the entropy of the source S. Therefore, the better code is when L(C) is closer to H(S).

H(S) <= L(C)

where the source S has the probability distribution as P(S)={p1, p2, p3,..,pn}
L(C)=Sum of {l1 x p1 + l2 x p2 + l3 x p3 + ...+ ln x pn}

How to improve our chance to achieve a career as an IT worker

The following helps us to improve our chance to work as an IT worker.

1. Good system designer
2. Good communicator
3. Good collaborator

How to improve our chance to achieve a career as an IT worker

The following helps us to improve our chance to work as an IT worker.

1. Good system designer
2. Good communicator
3. Good collaborator

What is computer's operating system?

1. Like a government to provide services for users (i.e. users' applications)
2. Like police to control traffic (e.g. I/O device controllers)
3. Like a facilitator between hardware machine code and application software
4. Provide networking stack for communications
5. Manage physical and virtual memory
6. Manage users
7. Schedule CPU(s) times for processes

What is computer's operating system?

1. Like a government to provide services for users (i.e. users' applications)
2. Like police to control traffic (e.g. I/O device controllers)
3. Like a facilitator between hardware machine code and application software
4. Provide networking stack for communications
5. Manage physical and virtual memory
6. Manage users
7. Schedule CPU(s) times for processes

Golden rules in chess

1. Attack the center
2. Look for danger
3. Move our fangs (i.e. knights or bishops) out early
4. Castle early

Golden rules in chess

1. Attack the center
2. Look for danger
3. Move our fangs (i.e. knights or bishops) out early
4. Castle early

Collaborative technologies

The following are some of current collaborative technologies

1. Instant messaging
2. Emails
3. Webcam
4. Voice over IP
5. Video conferencing
6. Blogs
7. Wikis
8. Videos

Collaborative technologies

The following are some of current collaborative technologies

1. Instant messaging
2. Emails
3. Webcam
4. Voice over IP
5. Video conferencing
6. Blogs
7. Wikis
8. Videos

Different ways to defend our attacked piece in chess

1. Move it out of danger
2. Protect it by another piece
3. Block it with a minor piece
4. Capture the attacking piece
5. Capture another piece that is more important than our attacked piece
6. Do a check or checkmate

Different ways to defend our attacked piece in chess

1. Move it out of danger
2. Protect it by another piece
3. Block it with a minor piece
4. Capture the attacking piece
5. Capture another piece that is more important than our attacked piece
6. Do a check or checkmate

Upcoming products

1. Thin TVs from Sony (about 2cm thickness).
2. Bluetooth TV (sending images from mobile phone to TV using Bluetooth connection).
3. DVD upscaling from SD to HD.
4. 3G IPhone problem between Wifi data and mobile network data.
5. TVs with Yahoo widgets (e.g. weather information, movies information, sport scores).

Upcoming products

1. Thin TVs from Sony (about 2cm thickness).
2. Bluetooth TV (sending images from mobile phone to TV using Bluetooth connection).
3. DVD upscaling from SD to HD.
4. 3G IPhone problem between Wifi data and mobile network data.
5. TVs with Yahoo widgets (e.g. weather information, movies information, sport scores).

What is knowledge?

Does knowledge mean

1. What you know?
2. What you really understand?
3. How much you really understand about something?

How knowledge is acquired

1. Reading books
2. Listening to radio
3. Watching TV
4. Attending schools and universities
5. Obtaining from our relatives and friends
6. Reading newspapers, magazines

Video lectures on electronics engineering

1. Basic Electronics at IIT Madras
   http://youtube.com/view_play_list?p=7987F30C41A9ADCB
2. Circuits and Electronics at MIT
   http://www.learnerstv.com/lectures.php?course=ltv024&cat=Engineering&page=1

Video lectures on electronics engineering

1. Basic Electronics at IIT Madras
   http://youtube.com/view_play_list?p=7987F30C41A9ADCB
2. Circuits and Electronics at MIT
   http://www.learnerstv.com/lectures.php?course=ltv024&cat=Engineering&page=1

Video lectures on algorithms and data structures

1. Algorithms at ArsDigita University
   http://www.aduni.org/courses/algorithms/index.php?view=cw
2. Introduction to Algorithms at MIT
   http://ocw.mit.edu/OcwWeb/Electrical-Engineering-and-Computer-Science/6-046JFall-2005/LectureNotes/index.htm
3. Data structures at UC Berkeley
   http://webcast.berkeley.edu/course_details.php?seriesid=1906978343
4. Theory of Computation at ArsDigita University
   http://www.aduni.org/courses/theory/index.php?view=cw
5. A Practical Guide to Data Structures and Algorithms Using Java at Washington University
   http://goldman.cse.wustl.edu/crc2007/lectures/
6. Data Structures, Algorithms, and Applications in Java at University of Florida
   http://www.cise.ufl.edu/~sahni/cop3530/index.html

Video lectures on Mathematics

1. Mathematical lectures using slide shows and videos at University of Colorado Denver
   http://www-math.cudenver.edu/~rbyrne/flash.htm
2. Linear algebra at MIT
   http://ocw.mit.edu/OcwWeb/Mathematics/18-06Spring-2005/VideoLectures/index.htm
   
   

Video lectures on Mathematics

1. Mathematical lectures using slide shows and videos at University of Colorado Denver
   http://www-math.cudenver.edu/~rbyrne/flash.htm
2. Linear algebra at MIT
   http://ocw.mit.edu/OcwWeb/Mathematics/18-06Spring-2005/VideoLectures/index.htm