Neural Networks explained in 60 seconds!
ฝัง
- เผยแพร่เมื่อ 21 ก.ค. 2022
- Ever wondered how the famous neural networks work? Let's quickly dive into the basics of Neural Networks, in less than 60 seconds!
▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT ▬▬▬▬▬▬▬▬▬▬▬▬
🖥️ Website: www.assemblyai.com
🐦 Twitter: / assemblyai
🦾 Discord: / discord
▶️ Subscribe: th-cam.com/users/AssemblyAI?...
🔥 We're hiring! Check our open roles: www.assemblyai.com/careers
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
#MachineLearning #DeepLearning #neuralnetworks
worst part about shorts is that if you miss something then you have to re-watch the whole short.
You can like it and it will go to the like section and you will be able to use the video player (timing)
you can scroll through time in shorts
There is literally a scrolling function embedded in the short 😂😂
You can seek the video
The worst thing is to believe that someone can teach a deep topic in 60 seconds by speaking fast, and people who believe that they will understand it. There are books to read or videos in detail to watch so you can pause to think, elaborate, sketch, write, …
This was the most impressive explanation of neural networks I've seen.
This taught me more than i learned about neural network in my previous life as a computer scientist
Wonderful short depiction however all of the magic happens in the back propagation/error correction phase .. it occurs to me that while the architecture is clear, any reasonably sufficient representation of a NN must spend at least some time discussing the training.
That's true! We plan to talk about backprop on its own video as a part2!
I think her explanation is wrong, I don't like when if you want to explain 'How does the ball bounce' every A.I guru's explanation is to explain newtonian physics behind. The explanation is much simpler. Can't even get full picture of why it works
@@AssemblyAI I appreciate your agreeing that you missed mentioning backprop.
Great vid! As always, the toughest talks are the shortest ones!
I don't subscribe to many but very grateful I did subscribe to Assembly AI ! Such a great breakdown
Thank you!
A simplified explanation but extremely good for beginners
Adding weights to inputs creates exponentially accentuates biases at the end results. So there is no escaping biases. Only a different type of bias
The best explanation of Neural Network. I have ever seen 😮 I am suprised ✨💯
Well explained 👍🏼
U r greaT Mam
Love From india keep sharing♥️✨
Some of the activation functions are :
1) Sigmoid function
2) ReLu
3) Leaky ReLu
What does this activation functions really do? Can u plzz explain?
Finally someone did it right. THANK YOU!
thankyou ☺️ please explain more concepts in long form videos
the last half video is quite good.
That was awesome. Instant sub.
Amazingggggg.....thanku.....finally....understood....!!!👍🏻
Very good explanation for just 60 seconds!
Another way to think about it is "it is just a mathematical function" and that is honestly scary to me. Are our brains just a gigantic mathematical function?
A neural network is just a giant mathematical function that takes multiple inputs and outputs multiple values. Training a neural network is giving the giant function some inputs, comparing the output with the desired output and updating the constants in the function accordingly.
I appreciate you comment as it interests me
Her explanation is how it works. It's very interesting. Your mathematical theory was a welcome message,
I could appreciate it if there was a numbered explanation,
I couldn't see it clearly, then looking at comments, I was trying to picture it.
Hearing her whilst reading your reply, I appreciated your insightful theory a matter of thinking numerically . Thank you for 🫴your comment.
Just a thought, when Ai, Chatbot4 etc are asked? Does it search a memory bank then go thru chosen descriptions for a particular type/style required to deliver 100% accuracy, similar to the animation exhibitions of a neural in action.
@@user-TimWill369 i dont think so
Say you are seeing a ball. The visual input is one. The sound is an other. The shape is another. Touch is an other. Bias closer to circles. Rolling. Sound rolling. Etc.
It is actually hard to explain such complex concepts in 60 seconds. Whether the viewer will understand it or not, depends from the knowledge he comes with.
am I too old to learn WITH NO education? Im 28.
@@WillisAmakyes it's over for you , your resume isn't even considered if you don't have a basic engineering degree and let alone you are too old
@@WillisAmak Definitely you are not old to start learning AI. The concepts are a bit hard to understand without proper math knowledge. However, it depends how hard you will work on it! Consider starting from basic Linear Algebra, Statistics and Calculus (Math for machine learning courses, for instance).
@@spectrohypernova8890bro 💀
Amazing explanation
Amazing explaation
What the short doesn't mention is that the weights and biases are updated using something called backpropagation which requires good knowledge of calculus to understand what is happening.
Mindblown ❤
saved my life !
Thank you
Very nice
Thanks
Stellar work! If you seek more, a book with like-minded content is what I'd propose. "From Bytes to Consciousness: A Comprehensive Guide to Artificial Intelligence" by Stuart Mills
Weights are optimised with a descent gradient, and the prediction is given by logistic regression, but how do you optimise the biases in a neural net?
So a neural network has a designed bias?
Designed inputs, as well as designed outputs.
What’s not controlled about it? The middle layer but still starts off (controlled/biased)
How’s it thinking? I’m dumb.
Merci 🙏
Can you clarify how do you handle errors at the end to mitigate them on the next iteration?
The use of x, y variable names is a bit misleading since x is often refered to as the input, while y, as output. Though if you are knowlegable about anns, it might not be a problem.
remember kids: always use reLU activation
Is this a pathway to quantum neural computers? As in, one which is robust to quantum uncertainty(thus doesn't need many forward passes, or uses multiple averaged outputs at once to speed it up), handles superposition within the hidden layers, and benefits from the exponential parallelization.
How does one interpret the output?
Watching this 15 minutes before exam 😭
I have no HS diploma, and im 28... What exactly are you studying? I want to learn this and know where to begin.
WOW 😳 amazing 😍 explain
Thanks! 😃
@@AssemblyAI ❤️🖤🤩🤩
have you ever seen a dual output layer in line?
0:00 for non-short form
My sis explained this in 60 seconds while my teacher couldn't in a semester
00:00
I need more info❤😮😮
To call this an "explanation" is like calling a firecracker a rocket that flies you to the moon.
Anyone link me something on how to do back propogation / good techniques
But why are hidden layers needed and how many of them?
Its seems pretty easy to understand 😂😅😂😂😂
Well explained. NN for Dummies.
What is the value of A ??
So do I integrate by parts? Or is there a trig identity??? 😂
😮
Well yeah, but that's missing the point. It's a matrix-vector transform in latent space.
60 seconds is not enough to understand what have been said in 60 seconds 😂
красивая девушка в современности
The music, tone and speed of speech, and the timer at the top are not really helping :/
For me it's her eyes.
Love from India 🇮🇳❤️
Love from American of Indian origin ❤️
Worst thing about this short is that you skipped a few steps...
im still confused [edit: i understand now}
All explanations of AÍ and neural network are very ambiguous and leave more image driven people behind. What is never answered is the nature of the neurons in a neural network. What are they in fact? Functions; concepts; math equations; transistors etc… like, in the brain is easy to picture the nature of neurons as a cell like structure exchanging neurotransmitters, but in the AÍ field it is harder to visualize what they rather than what they do. Maybe I should go ask chatgpt… 😅
It's helpful but still not totally clear. It would be perfect if there were a simple example, with simple values
xdd
Get explanation!
Is this some sort of competition of explaining this concept in 60 sec?. Why do you need a timer?
hahA
dada
Or, in one second.
it's complicated....
I don't see how you have explained them if you don't discuss backpropagation.
Could not understand anything. Is there any connection with real human body?
please dont include the time, because our brain will focus on the time and the discussion which is not good in your content.
I wonder what the purpose of such a video is.
how learining the english
Falls short You won’t get deep knowledge in 60 secs by no means sorry
The biggest issue is this is a dumb explanation