Neural Networks explained in 60 seconds! 

worst part about shorts is that if you miss something then you have to re-watch the whole short.

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.

I don't subscribe to many but very grateful I did subscribe to Assembly AI ! Such a great breakdown

Great vid! As always, the toughest talks are the shortest ones!

Adding weights to inputs creates exponentially accentuates biases at the end results. So there is no escaping biases. Only a different type of bias

A simplified explanation but extremely good for beginners

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

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.

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.

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.

saved my life !

Mindblown ❤

Thank you

Very nice

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.

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.

remember kids: always use reLU activation

How does one interpret the output?

Watching this 15 minutes before exam 😭

WOW 😳 amazing 😍 explain

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.

красивая девушка в современности

60 seconds is not enough to understand what have been said in 60 seconds 😂

The music, tone and speed of speech, and the timer at the top are not really helping :/

Love from India 🇮🇳❤️

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?

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