Видео

can u please provide dataset

How much durations should I consider for deep learning application in context of BCI MI-EEG ? One research paper used 4s duration for each events. However for each MI event how much duration is necessary to understand the brain response ?

Does retinal waves migrating from the temporal peripheral region of the retina to the optic nerve which would be closer to the nasal region?

Very good! Thank you for sharing.

Why he/him? Obviously a MICHAEL is a man xDDF

The Basal forebrain is n ot connected to the cortex at the first place. At adult hood you can see it as (NPD) Narcistic personality disorder) It's a link between the two!!!!

Great! Thank´s for making this video available. Thanks from Brazil.

法式英语比印度英语咖喱味还重

Why is there this strong assumption that PCs represents what motor cortex is trying to do?

Where on earth does one access this DystoniaNet platform? Been hearing about it for years but never a link to access it. How much does it cost? The neurologists don't know. The radiologists don't know.

How did you make allen_mouse_100um?

Intriguing.

So how does it actually "think" and "do" things?

Thanks. I wounder in slide 9, why there are many trajectories for the same condition if this is based on trial avergaed data. Shouldn't be there just one trajectory for each condition ?

Amazing study!!

Why not also do research also on Nithyanananda yoga? It's very active and authentic to Vedas and Agamas.

Mmmmm😊

Brilhante tema muita luz 🙏🏻Ⓜ️🕉️Om Nama shivaya Ⓜ️😇🙏🏻🔥💎♾️🗝️⚓️🌞👍🫵Ⓜ️Miguel Angelo .^.

Pure gold!

Well, a visual experience is not the entirety of imagination. If you're using your full imagination you know what you're experiencing is a work of imagination. If you're just having a completely internal visual experience then you're in what is akin to the non-lucid dream mode. Full imagination can be akin to lucid dreaming. But as I said visual experience is not necessary in full imagination mode. It can be very useful to not have visual experience turned up very high.

Great topic, great questions being raised, and great presentation but too many firm conclusions too soon.

Has someone tried using the G-Clusteron to join neurons in a neural network? I've looked online and I haven't found it. Would really appreciate it if someone could link to somewhere with it implemented or discussed.

Thanks for the video presentation!

How does the correction term help with stabilisation?

What about hyperphantasics who have a perfectly realistic imagination? For them, the strength model provides no explanation. Often people also describe their imagination as taking place on a kind of second screen. Isn’t that already a strong differentiating factor?

Why not Iunet directly to a single channel image? Why not golden angle radial sampling?

7:00

For anyone wanting a textual overview of the presentation - Michael Winding, a research associate at the University of Cambridge in the United Kingdom, presents his work on the complete insect connectome of a brain. He begins by posing a philosophical question: "If you had a comprehensive synaptic wiring diagram of a brain, would you be able to, based on that connectome alone, figure out the computations the brain performs to generate behaviors?" To answer this question, Winding and his colleagues chose the central nervous system (CNS) of the fruit fly larva. The CNS of the fruit fly larva was chosen due to its compact size, making it possible to quickly image using electron microscopy with synaptic resolution and reconstruct it. At the Janelia Research Campus, the team created an electron microscopy volume of the CNS. They have since been reconstructing the neuronal arbor and synaptic sites using an annotation tool called Catmaid. At this point, they have reconstructed about 3,000 neurons and about half a million synaptic sites. The connectome they've created allows them to follow sensory signals into the brain, through the computations that are performed there, and the behavioral outputs emitted by the descending neurons. They can even follow that signal to motor neurons which have been reconstructed in the nerve cord by many groups. Winding's talk covers the identification of cell types in the brain, how information flows through them, and some novel circuit motifs to give a sense of the kinds of analysis they can perform on this connectome. They've already identified all input neurons into the brain, and the next step was to identify output neurons. Michael Pancratz's group has already identified one type of output neuron called the ring gland neuron, which is responsible for hormone release. The team also identified and reconstructed all descending neurons to the brain stem equivalent and the spinal cord equivalent. These neurons control many different behaviors, including feeding, swallowing, orientation, navigation, locomotion, and nociceptive behaviors. The interneurons, the most numerous part of the brain, are key to understanding the computations the brain performs. The researchers broke these interneurons into chunks to better understand their roles, identifying many different categories of brain interneurons and three brain output types. In collaboration with groups at Johns Hopkins, they performed hierarchical spectral clustering to pull out a lot of structure from the connectome. They analyzed feed-forward and feedback signals between the clusters, discovering that feed-forward paths are multi-layered, with short, medium, and long paths. Feedback paths, on the other hand, are more robust across many brain clusters but become stronger towards the descending neurons. The researchers found that descending neurons emit feedback back to the brain, suggesting they may also emit an efference copy back into the brain, potentially as a teaching signal for things such as operant learning. By having both brain hemispheres reconstructed, they were able to identify all ipsilateral neurons, bilateral neurons, and contralateral neurons. They investigated the possibility of forming reciprocal loops between brain hemispheres and found that left-right homologous neuron pairs formed reciprocal loops with each other, specifically contralateral neurons. These loops could be involved in signal propagation, short-term memory, or inner-hemispheric signal comparison. In conclusion, the findings contribute to our understanding of the computations the brain performs to generate behaviors and pave the way for further research into the impact of genetics on brain wiring and neurological disorders.

17:00 cognitive building blocks

Interesting research. Too bad Mr. De Strooper acts as a manager with fear, resulting in his employees to be afraid to share negative data. Unfortunately, HR of KULeuven never acts on these issues until the moment this is shared in the media.

Is there any simple learning XOR gate example with equilibrium propagation?

I chose a very hard research article to analyze for my intro level Psychology course and now I need to become familiar with these types of models🤦🏾‍♂️. I’m a freshman with no prior experience on how these types of studies work and now I need to learn all of it just so I can understand what the research is showing. Pray 4 me.

Such ground breaking work and only 408 views. Very typical of todays society. 12:50 looks like a delay line. I think an electronic engineer would make sense of what is going on.

mio padre assoluto

Maybe I don't understand exactly what the "local extrema" are. But from what I understand, brain connectivity can be increased by many things, including musical training, reading complex works, using your non-dominant hand, meditation, etc. In contrast, IQ is more or less fixed upon reaching adulthood. If connectivity can increase/decrease without there being a corresponding change in IQ, then how can IQ be predicted from brain connectivity? Is it just that people who are already smart tend to more often do things that increase their connectivity?

Should try a density dependent repelling force between synapses

Poor the students!

where can we find the dataset?

What a fantastic speaker. This man is so thorough

Nice presentation.

☝🏿👆look up that handle, he ships swiftly, surest plug 🔌🍄🍫💨🤯

That was good

Thanks for sharing 💡

X-ray holographic nano-tomography would have been easier than electron microscopy for imaging

Just came to drop a comment and a like.

This lady first: "think FCN as only 1 neuron" this lady later: "I used a very implausable process called backprop" backward-propagating Action-Potential: "YOOO, THAT'S WHAT I DO! I TURN THE FACT THAT I FIRED INTO A TRAINING SIGNAL FOR MY SYNAPSES TO MAKE IT MORE LIKELY, YOOO"

Thanks for this really clear overview!!

Dataset link

Good explain

Interesting! What modelling techniques would you use to solve it? Time series like RNNs? or regression?