MIT 6.S191 (2020): Convolutional Neural Networks

Поделиться
HTML-код
  • Опубликовано: 17 янв 2025

Комментарии • 133

  • @mueez.mp4
    @mueez.mp4 4 года назад +153

    As a lecturer myself, I'm blown away by the quality of the slides.

    • @rito_ghosh
      @rito_ghosh 4 года назад +3

      True that, my friend. He knows how to create slides!

  • @beltusnkwawir2908
    @beltusnkwawir2908 4 года назад +39

    As a computer science graduate student, it's unbelievable how simple Alexander breaks down these concepts in such a way that I'm getting it from a fresh perspective

  • @vidyaverma6583
    @vidyaverma6583 4 года назад +9

    This is probably the best explanation of CNN that I have been through yet, it cleared almost all of my queries.

  • @domotorferenc
    @domotorferenc 4 года назад +13

    Congratulations! Very well organized lecture. High standard and easy to follow at the same time. I am a 68 years old mechanical engineer with average capabilities, but I enjoyed each minute watching and listening to your presentation. Well done! Thank you very much indeed.

    • @shacklemanwarts1305
      @shacklemanwarts1305 3 года назад +1

      I think that's why it's MIT. Usually I think of lecturer as scholar, someone with knowledge but not at the same time a good presenter, one of my lecturers spends most of class hours talking about how good he is, and not teaching.

    • @muhammadaarizmarzuq295
      @muhammadaarizmarzuq295 3 года назад

      @@shacklemanwarts1305 and just like that, he's gonna turn 69

  • @Cgodwinokolinta
    @Cgodwinokolinta Год назад +2

    Menhhh! This is arguably the best video on convolutional neural networks!!!

  • @reenietanya2289
    @reenietanya2289 2 года назад +2

    Amazing Lecture! The best on CNN. The only lecture probably that I’ve seen that cleared as to why each operations are done and how they influence the outcome. Thanks a ton. Much needed.

  • @lexifelix8510
    @lexifelix8510 3 года назад +10

    I never really understood convolution, but now I finally do! Thanks for this awesome lecture.

  • @FilipCodes
    @FilipCodes 3 года назад +1

    Wow, that explanation with the X in the matrix of 1 and -1, and how we look for the mini sub features and compose a bigger picture out of them was awesome.

  • @원형석-k3f
    @원형석-k3f 3 года назад

    19:02 Feature Extraction
    19:50 CNNs for classification

  • @adityavats9093
    @adityavats9093 3 года назад +1

    You cannot get a better lec for CNNs. Thank You MIT.

  • @gavinliu762
    @gavinliu762 3 года назад +3

    This lecture is so clear and powerful. I simply love it. Respect

  • @harshdevmurari007
    @harshdevmurari007 3 года назад +1

    just going to say...........it was mind blowing................hats off.......

  • @arthurkalikiti2856
    @arthurkalikiti2856 4 года назад +1

    By far one of the very best lectures of Computer vision.

  • @Justme-dk7vm
    @Justme-dk7vm 11 дней назад +1

    Handsome guy + Tech = Wholesome! Love from India!!!!!!!!!!

  • @impressivenow2000
    @impressivenow2000 4 года назад +2

    Hello Alexander, your presentation is very clean. I've seen many videos on CNN, and your presentation is actually very "clever" in the sense that you precisely tell about what's problematic and the reason why we solve issues this or this way. Pedagogically I admire your work, congrats.

  • @akshayjaryal8930
    @akshayjaryal8930 4 года назад +4

    Simplistic way to teach with awesome presentation contents. Thank you for your hardwork. Happy Coding !!

  • @mrworf314
    @mrworf314 3 года назад +1

    Best lecture on cNN that I have seen. Thanks mate!

  • @norbertfejer4015
    @norbertfejer4015 4 года назад +2

    Awesome. The best explanation I have ever seen. Congrats!

  • @qusayhamad7243
    @qusayhamad7243 4 года назад +4

    Awesome content and so clear explanation. Thank you very much for sharing these lectures with us.

  • @ishwarmorey712
    @ishwarmorey712 4 года назад +2

    Need to use a convolution is beautifully represented in image at 13:28.

  • @DarkLordAli95
    @DarkLordAli95 3 года назад +2

    This was amazing. Thank you Alexander!

  • @uchennanwosu5327
    @uchennanwosu5327 2 года назад

    Excellent exposition! The illustration of filters has helped me better understand the application of weights to learning. Please can you explain more about activation functions and their role with an example.

  • @Kppoja09
    @Kppoja09 4 года назад

    Clear and concise video on CNN. No need check another one on this topic. I am a student and performing plant disease detection and this would be very useful. Thanks !!

  • @abdullahbasit80
    @abdullahbasit80 4 года назад +2

    Wonderful, loved the explanations!

  • @tejaskhanna5155
    @tejaskhanna5155 4 года назад +8

    CNNs always intimidated me! This video made it really clear. 😀😀

    • @luisurena1770
      @luisurena1770 4 года назад +1

      I was firsr introduced to cnn's yesterday and i was super overwhelmed specially becausd i was introduced to it with a pre trained image classification model instead of breaking it down piece by piece like this, now i feel way better about it

  • @SR-de7gg
    @SR-de7gg 4 года назад +7

    Amazing lecture, thank you so much!

  • @amc8437
    @amc8437 3 года назад +1

    I know I will watch this several times over.

  • @muhammadsaadmansoor7777
    @muhammadsaadmansoor7777 4 года назад +1

    This is actually better lecture than that of Stafford I hope other lectures are available as well

  • @sdva3906
    @sdva3906 4 года назад +1

    Very useful video to start!

  • @욜로욜로-n1g
    @욜로욜로-n1g 3 года назад +1

    Thank you for sharing a great lecture!
    It helped me so much

  • @Popc1007
    @Popc1007 4 года назад +2

    Amazing lecture , thank you for sharing

  • @jatinkumar4410
    @jatinkumar4410 3 года назад +1

    Thanks. Very nicely explained.

  • @surya684
    @surya684 4 года назад

    Revisited all the basics. Great content!

  • @nottheengineer4957
    @nottheengineer4957 2 года назад +1

    Great presentation, but those bright blue slides with white text gave me a tiny shock every time they came up because I thought I was looking at a bluescreen.
    Was that intentional to keep people focused?

  • @Arslan-1
    @Arslan-1 4 года назад +2

    waiting anxiously :)

  • @markhopkins8731
    @markhopkins8731 4 года назад

    Amazing. Thank you Alexander.

  • @mugumyakevin8696
    @mugumyakevin8696 4 года назад +1

    Awesome content. Very intutive explanations.

  • @randomforrest9251
    @randomforrest9251 3 года назад +1

    Awesome Video! 20:28 doesn't pooling already bring in non-linearity?
    taking rectify for example. rectify(x)=max(0,x) leads to non-linearity like max pooling. right?
    Why exactly do we use pooling btw?
    Reasons I (think I) found in Goodfellow et at.:
    * whole regions are collapsed into one summary statistics, thus the network is more robust to noise
    * obv. we store less data and thus (most likely) need less parameters in the next layer (less time complexity..)
    Are there any additional reasons?

  • @zoumanakeita8016
    @zoumanakeita8016 4 года назад +8

    Great lecture, and thank you for sharing your knowledge!

  • @EBEALEX
    @EBEALEX 4 года назад +3

    Can't wait to see this😋😋

  • @jitendrakr9171
    @jitendrakr9171 4 года назад +1

    Thank you sir good Explaination of convolutional

  • @HiigsiTech
    @HiigsiTech 4 года назад

    thnks Alexander and MIT

  • @sarvatmir5888
    @sarvatmir5888 4 года назад +1

    Omg 😍mind blowing content

  • @shahriarhajiabbasi830
    @shahriarhajiabbasi830 4 года назад +11

    Feeling waiting in the dining room.... ready to have a big meal .... :)))

    • @ananyab4883
      @ananyab4883 4 года назад

      what

    • @shahriarhajiabbasi830
      @shahriarhajiabbasi830 4 года назад

      @@ananyab4883 That was for the time. We had waited for the class video to be uploaded :)

  • @isbestlizard
    @isbestlizard 4 года назад +2

    Pretty awesome that human brains can actually, in a split second, KNOW that 16x12 greyscale image was of abraham lincoln.. shame he didn't actually remark on that or invite the audience to guess who it was or anything..

    • @AAmini
      @AAmini  4 года назад +3

      Great point! The fact that I took this for granted demonstrates your point even further. But indeed, this is awesome :)

    • @isbestlizard
      @isbestlizard 4 года назад

      @@AAmini It is given there's only 192 pixels there, and humans know THOUSANDS of faces, and that it's an instant process not something that we consciously reason about, or iterate over. We definitely have abraham lincoln neurons in our brains XD

  • @kashyaptamakuwala3043
    @kashyaptamakuwala3043 4 года назад

    Great content and even a better explanation

  • @Dbeats0295
    @Dbeats0295 4 года назад +1

    yo this lecture is awesome

  • @developerhabits
    @developerhabits 4 года назад +1

    This is just amazing. The slides and the presentation are of such a high quality! As a complete beginner in the field, it's truly helpful.

  • @pedropereirapt
    @pedropereirapt 4 года назад +1

    Great lecture! Thanks for sharing

  • @firatbulak
    @firatbulak 3 года назад +1

    great lesson teacher thank you a lot.

  • @yunzhang8871
    @yunzhang8871 4 года назад

    Excellent lecture, so intuitive and easy to understand! 👍

  • @jm10oct
    @jm10oct 4 года назад +2

    Wait, where can I see the labs? I want to practice too

  • @kaifkohari5298
    @kaifkohari5298 4 года назад +2

    Great content make more videos like this 👍👍

  • @ivanparreno2342
    @ivanparreno2342 4 года назад +1

    one doubt, in minute 16:20, the filter has some numbers. Where dou you get this numbers from?

  • @varunjindal1520
    @varunjindal1520 4 года назад

    This is one-epic lecture on CNN

  • @DominoDay4Live
    @DominoDay4Live 4 года назад

    Why would you lose spatial information when flattening a fully connected layer? I understand that the amount of parameters to retain spatial information when using fully connected layers is massive, but spatial information is not lost when you flatten the data. The indices still relate to a region of the image.

  • @ruxiyan1603
    @ruxiyan1603 4 года назад +2

    waiting for the new updates!

  • @RrNefriana
    @RrNefriana 4 года назад +1

    I love you, Alexander! Thanks for sharing these videos with us!

  • @lifeisbeautifu1
    @lifeisbeautifu1 Год назад

    Amazing!

  • @katyhessni
    @katyhessni Год назад +1

    thanks sir

  • @tirabbit
    @tirabbit 4 года назад +1

    Great video

  • @mohsensafaei5289
    @mohsensafaei5289 3 года назад

    awesome, thanks a lot!

  • @appc23
    @appc23 3 года назад

    I have a question. How exactly the second convolutional layer is interfacing with the output of the first conv layer? Or, how is the filter of a layer-2 node slidden through the planes that make up the output of the first conv layer?

  • @owaisahussain
    @owaisahussain 4 года назад +1

    Awesome!

  • @shawnirwin6633
    @shawnirwin6633 4 года назад

    What is it about the negative numbers that makes the process want to reject them via the ReLu function? What is the reasoning behind rejecting those values?

  • @Wonder-g9d
    @Wonder-g9d 4 года назад

    Great it will help my project work thank u

  • @dimakilani286
    @dimakilani286 4 года назад

    Very informative. Thanks!

  • @MrRynRules
    @MrRynRules 4 года назад

    Thank you!

  • @trulyspinach
    @trulyspinach 4 года назад +1

    ahhhh, can't wait to see this

  • @abhinavrawal7183
    @abhinavrawal7183 3 года назад

    God bless you ...

  • @haarish58
    @haarish58 4 года назад

    How can we choose the number of filters in a CNN?
    by the way, thanks for the extraordinary explanation

  • @sagsolyukariasagi
    @sagsolyukariasagi 4 года назад +1

    15:49 How do we get that nine? when we perform multiplication it is not 9..

    • @puneetsingh5219
      @puneetsingh5219 4 года назад

      The '9' is achieved when you add all the elements of the output matrix [[1, 1, 1], [1, 1, 1], [1, 1, 1]].

  • @siddharthsharma9700
    @siddharthsharma9700 4 года назад

    Great video!

  • @James-hb8qu
    @James-hb8qu Год назад

    Neural Network: "Red light stop, green light go, yellow light go very fast."

  • @我想學英文
    @我想學英文 3 года назад

    21:04
    20:43

  • @sandeep832
    @sandeep832 4 года назад

    good one

  • @janakiramanbalachandran504
    @janakiramanbalachandran504 4 года назад +2

    Can you please clarify, if the RELU activation is applied individually on the outcome of every convolution operation for every filter. For example, if a 28x28 image is transformed into 26x26x3 convolution layer( with three separate 3x3 filters), the RELU activation is applied independently 2028 (26*26*3) times ?

  • @captainphysix
    @captainphysix 4 года назад +2

    nice

  • @mystery346
    @mystery346 4 года назад +1

    I lOVE IT

  • @mpaujan21
    @mpaujan21 4 года назад +2

    Awesome lecture! Thanks too, hopefully I can learn at MIT someday

  • @eitanas85
    @eitanas85 3 года назад

    Question: Does the filter weights are shared between all next layer neurons, or each of them has a different weights set?
    I tend to think that the second option is the right one, but am not sure

  • @alislamicity
    @alislamicity 4 года назад

    23:30 belum jelas ?

  • @julianr.starks6071
    @julianr.starks6071 4 года назад

    Can't wait to see this😋😋
    Great video!

  • @nimaramezannejad4243
    @nimaramezannejad4243 4 года назад +2

    great lecture and "Khaste Nabashi" :)

  • @hendmuhammad6238
    @hendmuhammad6238 4 года назад

    Thanks

  • @harunalperentoktas4983
    @harunalperentoktas4983 4 года назад +1

    Yeah!

  • @michpo1445
    @michpo1445 4 года назад

    Why does he keep saying connected to the hidden layer? Aren't the flattened images connected to the input layer, not the hidden layers?

    • @navdeepbeniwal1462
      @navdeepbeniwal1462 4 года назад

      The input layer for the dense neural network here is referred as the hidden layer, maybe since its not actually taking the input and has already started applying the weight right in that layer.

  • @kushangpatel983
    @kushangpatel983 4 года назад

    Very intuitive content! Thanks!

  • @zoumanakeita8016
    @zoumanakeita8016 4 года назад +1

    Lol. Waiting for the next great course!

  • @luissintra937
    @luissintra937 4 года назад +1

    Continuous label haha that was funny

  • @gundəgi-man
    @gundəgi-man 4 года назад

    Can anyone explain why there is additional dense layer in the classification ?
    I mean, the architecture for classification is composed with : Flatten->Dense(w/relu)->Dense(w/ softmax) .
    Why don't we just go : Flatten->Dense(w/softmax) ?

    • @hoaqyn8544
      @hoaqyn8544 4 года назад

      27:15 Listen to that slide again, especially at 27:46. The CNN extracts features from your image, but depending on the task, the use/importance of each feature is going to be different. That's the role of the dense layers in-between.

    • @gundəgi-man
      @gundəgi-man 4 года назад

      @@hoaqyn8544 thanks I've got the concept!

  • @jonathan364
    @jonathan364 4 года назад

    I was trying to understand your face recognition model and came acroos this line in the code: "loader = mdl.lab2.TrainingDatasetLoader(path_to_training_data)" What does this line do exactly. I see that it calls a package created by you guys. But what is it doing with the images?

    • @AAmini
      @AAmini  4 года назад +1

      This line simply loads the training images into memory. The actual code to feed the images through the model, train, test, visualize, etc comes later in the code. If you're interested, we open-sourced the entire MIT Deep Learning (mdl) package so you can see exactly what the internal calls do: github.com/aamini/introtodeeplearning/tree/master/mitdeeplearning

  • @jaidevshah2286
    @jaidevshah2286 4 года назад

    Great streamlined content

  • @emenikeanigbogu9368
    @emenikeanigbogu9368 4 года назад

    Finally. Letsss gooo

  • @jaredbeckwith
    @jaredbeckwith 4 года назад

    Learning how to do MNIST 🤖

  • @mfpears
    @mfpears 3 года назад

    The "convolution" component of CNNs is nothing more than a performance optimization.

  • @mahletalem
    @mahletalem 3 года назад +1

    💚💛❤

  • @marcinrutkowski9075
    @marcinrutkowski9075 4 года назад +6

    With such lecture quality I feel you could send a donkey to MIT and it would come out a scientist.

  • @trxstout706
    @trxstout706 2 года назад

    5:47

  • @ryuk-gj9uz
    @ryuk-gj9uz 4 года назад +1

    18:35