12a: Neural Nets

Share it with your friends Like

Thanks! Share it with your friends!

Close

*NOTE: These videos were recorded in Fall 2015 to update the Neural Nets portion of the class.
MIT 6.034 Artificial Intelligence, Fall 2010
View the complete course: http://ocw.mit.edu/6-034F10
Instructor: Patrick Winston

In this video, Prof. Winston introduces neural nets and back propagation.

License: Creative Commons BY-NC-SA
More information at http://ocw.mit.edu/terms
More courses at http://ocw.mit.edu

Comments

Max Johnson says:

Madagascar cat? Aren't those called lemurs?

Diana-Elena Ionescu says:

I lost 10 hours trying to understand the same thing from another set of lectures, such a waste of time! THIS MAN IS A GOD at explaining and you don't realise until you go somewhere else and get completely confused first.

Tarek Al Zihad says:

See, in 2015 mit professor is using windows 7. that's how you know windows 8 sucks

Álvaro Israel Nunes Leite says:

Thank you for this amazing class!

Paulo Constantino says:

25 years just to make a function continuous ? Are you serious ? That's the first thing I would do.

X X says:

It's sad that in our school we had lecture for this and I was lost but I think teacher was too. And than this guy comes with all elegance and no arrogance providing you this information and let it share too people around the world. WELL PLAYED.

Neel Basu says:

Why is it z(1-z) instead of P2(1-P2) ? in 38:50

Montserrat Cano says:

Thanks for sharing MIT! Excellent teacher!

David Parry says:

14:24…and right there, the flaw expresses itself.

David Parry says:

Excellent description of the Perceptron Model, devised in the 1950s, based on the 1943 work of McCullough and Pitts… 'tis a pity it's WRONG, can only be made to partially work using brute-force plus corrective mechanisms that simply don't exist in the actual tissue and the last near 70yrs of effort by countless AI-minions has been wasted on an unrepresentative model, based on a flawed interpretation of an incomplete bag of research based on non-human tissue that was dead. Luckily, we Europeans are on it, as usual. Onward to the future…

Bram Beer says:

Cool guy, awesome lecture!

Joseph Washington says:

2 years later and this is still a great lecture. Amazing instructor. I actually watched the whole thing. Simple ideas only take a quarter century to find. We humans need to make more observations, put them together, and see what shakes out.

jenny lennings says:

26:00

Close but no cigar lol

MinusBrain says:

Thanks for uploading such an awesome lecture. One point I did not get, though:
Could anyone please explain what i and j are in the function to calculate the delta of the weights at 21:24 ? Did I miss where the professor explains where this comes from?

idan hacmon says:

why did he choose to smooth out the step function in 25:00?

Jonathan says:

Is it me or does the professor really look like Donald Trump?

Chet Juall says:

Great ending beginning at 50:00

juan martinez says:

how many boards are there? lol

happy loser says:

Damn this guy is good! wow

happy loser says:

Dear Professor, Are you or is any one else here aware of P adic numbers being used to analyse the output of neural networks per Susskinds work on string theory? see below on an applied use of neural nets for landing fighters on aircraft carriers:

To realize quantitative forecast of carrier-based aircraft landing risk, this paper proposes a risk evaluation method implement the state description based on BP neural network, with """"Wave-Off Surplus Distance: WOSD"""" as reference index. According to the establishment of wave-off system with integrated control of military power and fuzzy elevator, the WOSD is defined with reference of wave-off envelope division for """"Ramp-Strike Risk"""". The burden of neural network training is reduced by establishment of """"State Risk Modeling Area: SRMA"""" based on limit wave-off envelope, finally the quantitative expression of """"Ramp-Strike Risk"""" in any flight states is realized through the design of BP neural network approaching risk evaluation function. Simulation results show that the outputs of BP network model designed basically accord with the expected ones, and the design of risk-evaluation method is feasible. This method can predict the """"Ramp-Strike Risk"""" in any flight states of carrier-based airplane, in addition provide early-warning and auxiliary rectification for safe landing.

Dao Xiong Teng says:

Good job with the cardio!

Panda says:

If you are blessed with such teacher, marry him/her.

Mike Schmit says:

Just a minor correction at 4 minutes.
That is a ring-tailed Lemur, not a Madagascar cat

Mustafa Ozturk says:

The lecturer is great, the camera operator sucks.

Daniyal Ali says:

learn a lot about neural nets from this video course.

Karthik Mohanarangam says:

At 28:00 why did he divide by 2? Is it because of two neurons?

Pratik Kulkarni says:

48:47 is that girl behind on the right side actually sleeping during such an interesting class?And that too resting her head on the girl next to her xD

ProjectPhysX says:

Put in a 1.5x and thank me later…

al Khwarismi says:

Great teacher. His heath doesn't seem so good, though.

Write a comment

*