Machine Intelligence - Lecture 7 (Clustering, k-means, SOM)

Поділитися
Вставка
  • Опубліковано 29 жов 2024

КОМЕНТАРІ • 32

  • @isabelmateus2547
    @isabelmateus2547 3 роки тому +77

    Intro to Clustering 0:27
    K-Means: 11:17
    SOM: 42:00

  • @iliasp4275
    @iliasp4275 3 роки тому +15

    Before coming here, i saw about 5 videos on SOM. No one pointed out that the algorithm is the same as K-means . You enlightened me! thank you very much

  • @Simzreid
    @Simzreid 4 роки тому +11

    Absolutely brilliant. Clear, concise, flowing and enlightening. A great help in understanding SOFMs research I am thinking about doing. 👍

  • @intoeleven
    @intoeleven 4 роки тому +52

    42:00 starts to talk about SOM

  • @subramaniamsrivatsa2719
    @subramaniamsrivatsa2719 4 роки тому +5

    My humble regards to Professor. For so much simplifying complex concepts and explaining intution behind the algorithms ...and encouraging us to understand 🙏

  • @ravivarma5703
    @ravivarma5703 5 років тому +48

    this is the best explanation i have ever found. Please is there any way i can see more lectures from this professor in any other channels?

  • @rezamonadi4282
    @rezamonadi4282 3 роки тому +5

    I am really proud of you! You explained SOM like an exciting journey...

  • @MrQqqq2222
    @MrQqqq2222 4 роки тому +6

    Wonderful professor. I can follow with him even if i am so far from ML field. I start to love Mr hamid and also AI methods and techniques. Thanks a lot my favorite virtual teacher.

  • @joecrowley630
    @joecrowley630 5 років тому +7

    So good, loved the enthusiasm and A-class white board usage of the lecturer. Thank you so much for sharing.

  • @derollo3
    @derollo3 4 роки тому +5

    Excellent Lecture about clustering. Thank you very much for sharing your knowledge.

  • @ab8jeh
    @ab8jeh 5 років тому +4

    I like the way he explains things very clearly. Within machine learning there is a tendency to cloud things to make oneself seem more intelligent - this lecturer shows how simple some of these algorithms (and ML in general) truly are without dumbing things down.

  • @judealaitani1036
    @judealaitani1036 4 роки тому +7

    Amaaaaazing teaching skills!

  • @karannchew2534
    @karannchew2534 Рік тому

    SOM 40:39
    1:14:25
    Given input X, find i-th unit with closest weight vector by competition.
    WiT X will be maximum.
    Find the most similar unit.
    i(X) = arg max i Ⅱ X - Wk Ⅱ
    k = 1, 2, 3... m, m = no. of units.
    The "max" here means highest value of dot product. The most "aligned" set of vectors between input vector and the neuron vector. If the vector are misaligned, the dot product (think cos θ) might be zero.

  • @isuruvindula2346
    @isuruvindula2346 4 роки тому +2

    Sir, you are a lifesaver sir. Thank you very much.

  • @karannchew2534
    @karannchew2534 Рік тому

    Given input X, find i-th unit with closest weight vector by competition.
    WiT X will be maximum.
    Find the most similar unit.
    i(X) = arg max i Ⅱ X - Wk Ⅱ
    k = 1, 2, 3... m, m = no. of units.
    The "max" here means highest value of dot product. The most "aligned" set of vectors between input vector and the neuron vector. If the vector are misaligned, the dot product (think cos θ) might be zero.

  • @Birdsneverfly
    @Birdsneverfly 4 роки тому +3

    Let me take a moment to admire your handwriting :) Plus "You are becoming your data", this has to be a dialogue from an AI movie. Cheers :)

  • @karannchew2534
    @karannchew2534 Рік тому

    Very neat handwriting for a professor.

  • @Mark-wb8ck
    @Mark-wb8ck 4 роки тому +6

    SOM starts at 40:42

  • @burakkara337
    @burakkara337 6 місяців тому

    Professor says, "Nobody screams when I make mistakes". I went crazy on monitor nobody hears me :D 28:32

  • @jm-px3mr
    @jm-px3mr 3 роки тому

    Thanks for sharing the amazing lecture. I wish to take a class in real someday

  • @mohammadghorbani188
    @mohammadghorbani188 5 років тому

    Very good lecture. Thanks for sharing.

  • @homataha5626
    @homataha5626 4 роки тому +3

    does anyone know implementation of SOM in python 3?
    in all the code I have seen they always use targets but we don't have it what should we do?

    • @pechhase
      @pechhase 4 роки тому +1

      you could check Peter Wittek's somoclu library

  • @basics7930
    @basics7930 4 роки тому

    great explanation

  • @sara-ja
    @sara-ja 4 роки тому

    perfect ...thanks

  • @venkateswaranvenkatraman9630
    @venkateswaranvenkatraman9630 4 роки тому

    why people are taking Mahalano distance , then eculidian distance ?

  • @yangzuo6973
    @yangzuo6973 4 роки тому +2

    Who is the other scientist who has got new ideas?

    • @Reggelord
      @Reggelord 4 роки тому +2

      "Norbert Wiener and Albert Einstein"

  • @yeyez3149
    @yeyez3149 4 роки тому

    lecture notes?