Perceptrons V Neurons - Machine Learning Is Not Like Your Brain #2

Рет қаралды 7,390

Күн бұрын

Пікірлер: 18

@HD-Grand-Scheme-Unfolds 2 жыл бұрын

I love this series. It was time someone took up this topic. To compare the differences between natural and artificial neuron (perceptron). The answer may be in the differences to be abstracted. Spike timming (in order of frequency or arbitrary rhythm) and phase. Not just importance emphasised only on spike intensity magnitude. (Which it worth noting that spike intensity is analogous to perceptron output value in ANNs, after operation by weight and bias has already transpired, an right before summation phase begins in the next step.) These values play roles in fundamental principles of how control and emergent intelligence-based phenomena manifest into reality. They add more computational dementionality and parallelization and task-performance-multimodality. Nice school of thought that you are presenting in saying ANN need to catch up more in its versatility to natural neurons 👍👍👍👍👍.

@darkninja___ 9 ай бұрын

I am always trying to learn and understand more about myself, the world, the brain, and most importantly consciousness and cognition itself so it’s cool to see someone making content like this. You deserve more views. Ty for the videos.

@mrknesiah Жыл бұрын

Beautiful video. Thank you. This exactly addresses the point of misunderstanding about "AI." This is one reason why a bee can have intention, solve problems and navigate in a time-sensitive 3D environment using 0.002 watts.

@veda_dom 2 жыл бұрын

Very informative videos. Thank you for creating this series, looking forward for more videos.

@akompsupport 2 жыл бұрын

Good explanation of perceptron concept.

@sourcesource8750 2 жыл бұрын

You know how much data our senses daily collect, it also is similar to how deep neural networks work

@FutureAISociety 2 жыл бұрын

Thanks for your comment. I'll delve more into this topic in an upcoming video in this series. You senses _receive_ a lot of data (the optic nerve alone is about 1 megabit/s) but actually _collect_ only a tiny fraction. The way that your brain receives the information and decides what to keep and what to discard has very little relationship with a deep neural network.

@applepie7282 2 жыл бұрын

@@FutureAISociety I think deep recursive neural network model solving this problem. the hidden state of an artificial neuron working like intrinsic memory mechanism of a biological neuron if Im not wrong

@rumyhumy 9 ай бұрын

Underrated

@paganaye Жыл бұрын

At 2:55 you mention a negative weight. I thought that there is not such thing as a negative spike in the biological brain. Does it represent a lack of spikes?

@FutureAISociety Жыл бұрын

Excellent observation! In biological brains, there are different "flavors" of synapses. Some contribute positive ions which (when accumulated) lead to the creation of an output spike. Other contribute negative neurotransmitters which inhibit the creation of a spike. In computers, we can emulate this by simply creating a negative synapse weight which does the inhibition and the weight can be modified smoothly from positive to negative but in biology, excitatory and inhibitory synapses are distinct from one another because they contribute different neurotransmitters.

@primodernious Жыл бұрын

ball neuron i think he mean that there is one node that activate to output the sum of a whole lot of nodes when the network detect a ball.

@secretname2670 Жыл бұрын

it's really simple if you use simple words; humans imagine , transform and conceptualize AI identifies, copies and merges the more examples of work done by both we receive, the more apparent that becomes and this truly is why at the peak of human ability they are better than AI but at the low they can't even match it.