I have come across interesting presentation about new chips which are being develop or appearing on the market. Those new chips are having NN like architecture and are design to be used instead of current CPUs where the computation and power consumption requirements to complete the tasks are high.
1) Do you think we will build one day a chip which has the same amount of neurons than human brain?
2) What field of application do you think will gain the most from this new chip development?
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
Thanks
1) Do you think we will build one day a chip which has the same amount of neurons than human brain?
ReplyDeleteI have learned to never say never. Today it seems not possible, but with the development of technology I would say that one day it might happen. I was interested in field of neuro-informatics some time ago, and there is a lot that majority of people have no clue about. With more and more specialists in this area I hope that we can develop more than expected.
2) What field of application do you think will gain the most from this new chip development?
I think that the whole computers will change in terms of architecture and usage. We want more and more from our hardware. This will be a revolution in all areas of life starting from science till business. Possible usage example might be hardware that can distinguish whether food is no longer fresh based on look, smell and temperature. The other example are smart traffic lights that based on image, sound and smell will detect possible danger of accident.
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
I think that normally people working in office or just in home will not need such technology. In the other hand using such advanced technique might be to complicated for them. So in my opinion they will work in parallel accordingly to the task at hand.
Thank you Cezary, I think we will just add additional unit like we have GPU and CPU right now we will have NPU and our software will use NPU for specific tasks.
Delete1) Do you think we will build one day a chip which has the same amount of neurons than human brain?
ReplyDelete‘One day’ can be really distant future, but I think this is possible. What I think will be developed earlier will be chip with less number of neurons than human brain but with similar capabilities.
2) What field of application do you think will gain the most from this new chip development?
It can have a lot of different applications, voice and video recognition, natural language processing, recommendation systems etc. basically anything that right now is solvable by deep learning.
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
It will be rather used in parallel. It is because this kind of chips are rather not good general purpose device and there will be still market demand for such devices. At this moment for instance CPU are still in common usage but GPU are developed as well
Thank you Ewelina,
DeleteYes sorry for not being specific about exact time, I've been watching few years ago Ray Kurzweil https://www.youtube.com/watch?v=1uIzS1uCOcE talking about Singularity. So might be those chips will contribute to his thesis.
Thanks, it was a very interesting presentation.
ReplyDelete1) Do you think we will build one day a chip which has the same amount of neurons than human brain?
It seems to be possible. However it depends on the speed of technology development. Maybe it wouldn’t be a one chip but a set of many chips responsible for specific tasks. Furthermore I think it won’t be so efficient as human brain is but in spite of everything it will be a highly advanced Artificial Intelligence.
2) What field of application do you think will gain the most from this new chip development?
As it was told in the presentation - everything that needs to be personalised: mobiles, computers, smart homes, cars etc. In my opinion it could include a huge area of our everyday life. Maybe such chips will replace current Central Processor Units in computers.
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
In my point of view firstly new chips will be used parallel to current components at the very beginning. Secondly I think that only new chips will be produced because of their performance advantage. It can be an exciting technology that we will feel in every-day reality.
Thank you Damian, I guess we will see. What amaze me there is such an advance in this field and there is no much publicity really about it. I guess it's because there are advancements in many fields each day and people get used to it. My point is I don't see it as an advancement but rather big change in approach.
DeleteAnyway thank you for your answers.
M
1) Do you think we will build one day a chip which has the same amount of neurons than human brain?
ReplyDeleteOne day... Do you mean in next century, millennium or decade? Human brain is quite complicated and from what I know we are not even close. It's hard to say that we are in the middle of research. I can't imagine computing power of such chip.
2) What field of application do you think will gain the most from this new chip development?
Bio-Chemistry, Medicine, everything? Weather predictions could be more accurate and so on. Maybe we will be able to build really autonomous transportation? Who knows?
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
We have a lot of different architectures now. Most of them are built for certain scenarios. However, our current cpu architecture is quite old and something like this would be a big break through. Even though, it will be hard to introduce it. It's really hard to predict something like this.
1) well as per presentation they are saying within next 2 decades we should be able to do such a chip so it's not really so far
DeleteThank you for your comments Mateusz
1) Do you think we will build one day a chip which has the same amount of neurons than human brain?
ReplyDeleteLet's think for a while is this massive process is possible in distant future. We are building faster computers each year that require more amount of power however healthy human brain consumes hardly 10W. I don't question if this finalize one day but I don't see this quickly happen.
2) What field of application do you think will gain the most from this new chip development?
All entertaining application surely would profit in this progress. I do not know perfectly where development going to start.
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
I am not an enthusiast of making human brain chips parallel. Can you imagine that your legs can run without any synchronization without your brain process?
1) Well the new chips are consuming 1000 times less energy because of the new architecture and this is just the beginning of their development.
Delete3) I think you haven't got my question right, I was talking about old chips and new chips running next to each other
Anyway thank you for your feedback.
M
1) Do you think we will build one day a chip which has the same amount of neurons than human brain?
ReplyDeleteI don't think I am capable of doing that, no.
Getting serious, given the required complexity of such Integrated Circuit, I am not sure whether it will happen during my lifetime.
2) What field of application do you think will gain the most from this new chip development?
I am not a futurist, I can't predict applications of (in fact) future technology. It depends on so many factors, that I don't even want to guess.
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
There are many possible computer architectures, let alone electronic ones. We may live to see many realizations of IC's used in parallel, i.e. quantum optical computers (regardless of their internal structure), traditional silicon ones, superconducting chips, etc. I can't really predict which of these would gain mass adoption.
2) Well presentation pretty much stated what field of application it's currently used with new chips I think any field which need NN will profit from those chips.
DeleteThanks for your answers
M
1) Do you think we will build one day a chip which has the same amount of neurons than human brain?
ReplyDeleteYes, in my opinion it is possible. Not in a near future but one day I am sure we will be able to do that. It depends just on technology we will be using and now it is growing really fast.
2) What field of application do you think will gain the most from this new chip development?
Maybe some medical application, mobile phones, computers... Everywhere where we are using chips currently and more.
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
I think new chips won't take place of current chips. I do not know if they will be compatible with devices we are using so they will work rather parallel.
Thank you Katarzyna.
Delete1) Do you think we will build one day a chip which has the same amount of neurons than human brain?
ReplyDeleteThe brain is a deviously complex biological computing device that even the fastest supercomputers in the world fail to emulate. Well, that’s not entirely true anymore. Using the NEST software framework, the team succeeded in creating an artificial neural network of 1.73 billion nerve cells connected by 10.4 trillion synapses. While impressive, this is only a fraction of the neurons every human brain contains. Knowing this, it shouldn’t come as a surprise that the researchers were not able to simulate the brain’s activity in real time. It took 40 minutes with the combined muscle of 82,944 processors in K computer to get just 1 second of biological brain processing time. While running, the simulation ate up about 1PB of system memory as each synapse was modelled individually. Computing power will continue to ramp up while transistors scale down, which could make true neural simulations possible in real time with supercomputers. Eventually scientists without access to one of the speediest machines in the world will be able to use cluster computing to accomplish similar feats. Maybe one day a single home computer will be capable of the same thing.
2) What field of application do you think will gain the most from this new chip development?
If you need an answer and you need it fast, ask our Crystal Ball for a simple yes-or-no solution. Going back to basics it is very difficult to predict if that development will have the real usage or it will be the dead end.
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
Please see my answer for the second question. On the begin it will work in parallel if it will work at all. The future in unpredictable.
1) Well nice feedback what is K computer?
Delete3) I'm sorry but those chips are already being used so it's not really the Crystal Ball question.
Thanks for your feedback really like the 1st answer.
M
1) Do you think we will build one day a chip which has the same amount of neurons than human brain?
ReplyDeleteI have no idea. Maybe scientists who are reaserching this particular field will create it, but I don't know how much money it can take, how much effort it can take or how many scientist should work on it to create it in finite time. Of course, I can see many advantages of an artificial brain or a machine which would have brain's computing capabilities. So, to sum up, I think that maybe it will be created someday, it will be an enormous achievement for the science.
2) What field of application do you think will gain the most from this new chip development?
I agree that maybe medical field could benefit the most, because decision engine would work much faster and would be more precise, as it would be deprived of emotions.
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
It is very hard to predict indeed. I think they will work in parallel and for sure they won't replace the current architecture in short time after they will be created.
Thanks Pawel, I think if those chips will gain some traction in business solutions they will just got their investments in place.
DeleteThank you for your comments I hope you like the video.
M
It is looks like no one watches the movie before answering the questions :)
ReplyDeleteI will answer them according to Henry Markram presentation.
1) Do you think we will build one day a chip which has the same amount of neurons than human brain?
Well according to Mr Markram technology he speaks about can handle a trillion neurons. It is more than in human brain. Problem is that they have only one input to each neuron so it is actually go to about a billion neurons with close to hundred billion synapses. As we go further in next ten years we will have the hardware to capture the scale of the numbers that are in the brain. So it is quite obvious that it is a matter of time as we reach the number of neurons in human brain. The number is not a problem. Programming this will be the most difficult challenge.
2) What field of application do you think will gain the most from this new chip development?
The most useful usability is the data interpretation. Today we generate so much data that we can not interpret them in enough short time. This new neuron based chips are design to handle lots of data.
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
It depends of developer. For example the IBM company is working on the stand alone neuron chip but Qualcom is working on this as a part of a everyday CPU. According to them future processors will have NPU (neuron processor unit) next to regular CPU ans GPU and other parts of mobile processors.
1) Yes the only correct answer to specific question as I didn't really ask about capabilities of the new chip or if it will be better than human brain lol. Just asked about numbers of neurons which as you noted will happend=) Maybe presentation was too long ...
Delete3) The same as point 1)
Thank you Rafal really appreciate watching the video=)
M
Your Welcome :)
DeleteIt was quite refreshing presentation. Thanks to this I put this subject to my reminder as a off topic research. Just because I am curious what are they doing in this filed right now.
Hey thanks for your video recommendation. As for the questions:
ReplyDelete1. This will surely happen one day, some visionaries predict it could happen before 2050. Let's wait and see or even better, work to help the cause. ;)
2. Mostly those that are heavy computation reliant, anything that is deemed to be intelligent enough to work autonomously. Other fields could be security breaching or prime number searching. ;)
3. At first they will probably be used together with current solutions but with time they might exceed the current standard architecture capabilities and become the new standard, totally replacing it.
Thanks Michail.
Delete1) Do you think we will build one day a chip which has the same amount of neurons than human brain?
ReplyDeleteI think this is possible and is just a matter of time. Technological progress is very difficult to predict with certainty, same as scientist cannot predict the result of experiment (what would be the point of conducting the experiment if the result is already known???). Looking on the world these days and even 10 years ago, I am of the opinion that chip with same number of neurons as humans can be invented sooner than we think.
2) What field of application do you think will gain the most from this new chip development?
Primarily banking, data scientists, manufacturing, - pretty much all areas where human can be replaced by machine.
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
Probably the new chips won't take place of current chips. It’s still difficult to talk about the future in this field.
Thank you Mateusz for your answers, I also agree it's fairly possible within reasonable time only if we will find business application behind it.
DeleteM
I’m not sure that we will be able to build such chip BUT if I’m wrong that can be the end of our time as the most important specious on our planet. If any machine will have similar “CPU” than we have than we`re doomed.
ReplyDeleteSuch powerful chip would be very useful in medicine and astronomy. Analyzing very big amount of data with some context is a very power consuming task.
First of all why do a typical person need such technology ? It would just increase the cost of our hardware so this is not a good idea. Secondly I don’t think that this would be even possible. We are getting already to the limits of the current architecture. So it probably will be (and should be) a parallel solution
Lol, come one we can always turn off the lights=)
Delete1) Do you think we will build one day a chip which has the same amount of neurons than human brain?
ReplyDeleteOn one hand, it seems to me that such possibility is real if the question that we pose is only the amount of neurons in neural net. On the other hand, neural net model is still drastically simplistic when compared to wonders of human brain. It will take decades of research to build a more 'real' illustration of human thought process. With new successes in AI, multiple articles and news published every day it is tempting to think that we are getting close to cracking human mind and thought. However one should not forget that the problems that machine learning is solving nowadays are still toddler level on complexity scale when you compare it to what an average human can do.
2) What field of application do you think will gain the most from this new chip development?
Such chip would mean an astonishing increase in computational power, thus all fields relying on computational power would gain a lot. As mentioned, all neural nets applications would increase significantly in accuracy, due to practically unlimited possibilities of training data scale - it would be possible to feed models with unimaginable amounts of image/text data.
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
If the new chips one day were as powerful as expected their popularity would mainly depend on the manufacturing cost. If it was possible to make it relatively low then of course new chips would be made a new standard and there would be no reason to use the old ones.
Thank you Ivona for elaborate answer. I would argue about point 3 as new chips are really designed for different tasks solving than current chips.
DeleteThank you for extensive answer.
M
1) Do you think we will build one day a chip which has the same amount of neurons than human brain?
ReplyDeleteWhy not? Future will bring new possibilities and solutions we have not thought about earlier. Everything is possible. The real question is: Do we want to continue the research? What is the desired effect? What do we want to do with it?
On the other hand, human does not use the full capability of it's brain so it seems to be reasonable to start building individual modules bit by bit.
2) What field of application do you think will gain the most from this new chip development?
Any activities engaging transformation of data into information. To be more specific - translations, voice recognition, etc.
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
It depends on a specific task. Some should be processed in patallel some will not benefit from parallelization.
1) Well I think the main goal is to design it for all NN related tasks where old architectures need really a lot of energy to calculate reasonable result.
Delete3) I was rather asking about if by building new chips we will be able to get rid current architectures or if we will still use the old ones for some purposes and new one only for those NN related.
Thank you for your feedback.
M
1) Do you think we will build one day a chip which has the same amount of neurons than human brain?
ReplyDeleteNo we want build such chip in near future.
2) What field of application do you think will gain the most from this new chip development?
Everything connected with AI and human feelings recognition.
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
Chip architecture is constantly changing and AFAIK now every Intel new processor is made in different architecture but when it comes to this one I think that standard chip will work in pararell with this one.
3) Yes well but we are talking here about totally new design but yes i think we will just get additional NPU like we have GPU right now.
DeleteThank you for an interesting presentation.
ReplyDelete1. I think that everyting is possible. Bearing in mind the speed of technological progress I guess it is a matter of years. Neuro-informatics is a discipline with huge potential.
2. It could be applied in so many field of science: medicine, neurology, linguistics, but also in applications we use on a day-to day basis.
3. I agree with my previous speakers. Such chips might be used parallel, as there aren't that many professions, jobs, companies that require such advanced pieces of technology to be working on a decent level of efficiency.
1) I totally agree
Delete2) I think we will build and understand a lot more about human brains along development of those chips so probably we will even find new fields of applications
3) OK
Thank you,
M
1) Do you think we will build one day a chip which has the same amount of neurons than human brain?
ReplyDeleteYes. Why not? The technology is developing really fast. But I am not sure if the amount of the neurons is a main topic. Human brain is really complicated structure, that can learn and function in very flexible way, so maybe such ability to change will be the biggest challenge in future computing.
2) What field of application do you think will gain the most from this new chip development?
The ones that processed a lot of data and need a good performance, like face recognition or medical purposes, like analysing the results of complicated medical test results (like MRI).
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
It is a good question. Probably it will work in both situations, but as it happens with new technologies, at the beginning it might be used as parallel solution.
1) I guess we need to do it one step at the time. For sure on the way of developing more and more neurones in on chip we will also build the whole new branch of computer science knowledge
Delete3) From what I understand for the moment we will get this chip as additional chip we will see what will happen next.
Thank you Emilia
We are not able to create such CHIP …. Yet ;). I am not sure will we ever be able to create something what will be as efficient as human brain. Please take a look on vision systems. Their are doing very precise job but in specified environment. We are supporting this systems by stabilizing environmental conditions (lightness, distance etc.) Humans' brain can adjust to environment very fast and efficiently. If we are talking about CHIPS architectures I think that the TOP manufacturers like Intel always works in parallel.
ReplyDeleteThank you for your input Pawel
Delete1) Do you think we will one day build a chip Which has the same amount of neurons than human brain?
ReplyDeleteSome day for sure. Let's see Moore Law. I believe that this is possible
2) What field of application do you think will gain the most from this new chip development?
I think that will benefit All areas. But first research solutions, perhaps the military?
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
Unfortunately, I do not know too much about it.
2) Lol, I think it pretty much came up from some Military contract=)
DeleteThank you Maciej
Thank you for the very interesting material.
ReplyDelete1. I think it is possible - in fact it is much harder to say what is not possible. However, in my opinion, it has to be another very long time that humanity created processor par with the possibilities of the human brain in one day.
2. I agree with previous speakers that it is being a new class and a completely new possibilities that can be successfully used in a variety of intelligent devices, significantly improving their functioning.
3. Initially, they will work in parallel, then they replace the current solutions. This is the safest way to introduce a variety of new products. If you create a better life then they displace the old solution.
Yes I think it's possible as well and probably not in so distant future as some may predict.
Delete1. That sounds plausible. But I am afraid that not In the nearest future. I wouldn’t be exuberant about it but we should always keep on trying.
ReplyDelete2. In a lot of spheres. It is hard to mention all of them. It seems to me that neuron based chips can have a wide usage.
3. The new architecture can be initially based on the former one and take a lot from them but with time it can be developed to a greater extent.
Thank you ZC
Delete1) Do you think we will one day build a chip Which has the same amount of neurons than human brain?
ReplyDeleteSome day for sure. Let's see Moore Law. I believe that this is possible
2) What field of application do you think will gain the most from this new chip development?
I think that will benefit All areas. But first research solutions, perhaps the military?
3) Can new chips take place of current chips architecture or rather they will work in parallel being used accordingly to the task at hand?
Unfortunately, I do not know too much about it.
goldenslot
gclub online
I found it interesting how your notes only focus on website..
ReplyDeletedoctor strange (2016)