Anyone with proper Win7/Win8.x license will probably be able to sue M$ for this. Since their technical support is still active and the hardware at setup time was accepted as compatible, this is a violation of EULA. It is like buying a car and then the company suddenly changes the left-side steering wheel with a right-side one, mandatory to keep its service active. Typical M$. Switch to Linux now.
Great. Can now Nestle do the same with cocoa, so that children in Africa do not have to work 16 hours daily for $2 in order for the company to make the chocolate? Google's AI promos are plain stupid, but this is gross. This science-for-marketing thing has gone too far.
The petition has already been proven a joke. It was hijacked by bots all over the globe. I am surprised that the UK gov actually made a statement about this. Well, maybe not surprised at all...
Earth circumference is about 40075 km associated with 360 degrees of angle (lat,lon). This means at most 111.32 km per degree, i.e., less than 31 m per angular sec. If we use GPS coords and two typical 'float' (32-bit) with epsilon accuracy at least 10e-5 we can have 5-digit decimals or roughly 1.2 m resolution, which is far smaller than the 5-8 m maximum resolution of commercial GPS.
So, with just 8 bytes we can have more than enough location registration anywhere on the Earth with the maximum available technology today. Unless these guys have found a way to pack 3 readable words in just 8 chars with no collisions whatsoever, they deserve the next Turing award - or the "bubble of the year" award in any other case.
The local Afghan governor said yesterday that "sadly, they had to to di", because Talibans were passing through, getting treatment from MSF and borrowing (temporary stealing) the cars. Also, there is no other airforce in the area and no other hospital in that region, located ***between*** two Taliban positions (see map).
It was a clear-cut tactical decision to destroy the hospital and they kept on bombing it for almost a full hour until the central building was leveled to the ground. The MSF central has already announced that they are pulling out of Kunduz completely, which was in fact the real goal behind the bombing. If they go back there, they will be bombed again, and again, and again...
The author has already given full-access licence with the latest version of the software, non-revokable, back in 2011. If an update is scheduled, only THEN he can alter the distribution and usage selectively.
Furthermore, there is no legal way to back this up, since the international trieaties forbid any customer discrimination based on race, religion, political views, etc. This means that someone WILL use his software and if he decides to prosecute him/her, the court will drop the charges.
In practice, there are several thousands of scientist in other countries that will be willing to help. Send them your scripts and they will run them for you. No one can bring racism into science. Period.
Seriously, this is perhaps the most ridiculous "study" ever from OECD. It is nothing less than blaming the use of pens instead of pencils for bad exam grades.
Someone should eventually get in the media and apologize for this as a hoax. And do it fast.
1. Yes, the brain is massively parallel and "analog" - BUT not every neuron forms a distinct cognitive function (FBNs) and neuron do fire in pulses/spikes (almost binary) rather than continuous (analog) outputs.
2. No, the article does NOT identify 'cpu cores' in the brain. It uses this metaphor (stated clearly in the paper as such) to point out the level of parallelism needed to run anything remotely similar to the complete functional 'package' in the brain.
3. The resolution of modern fMRI is at 3mm^3 (30K-50K active voxels) but this has to do more with the localization of the activations and less with the inherent complexity (dimensionality) of the spanned data space. In other words, in this work e.g. the visual center is detected as activated or not, regardless of how fine the resolution is.
4. The fMRI captures the complete 3-D brain volume, hence the detected activations include all the "always on" circuitry like respiration, cardiac rhythm, etc. Cognitive processes are only a few of these activations and are identified by experts when looking at the actual activation maps.
5. The methodology is completely data-driven and it includes two very popular non-parametric approaches: one is ICA for blind-source separation (measuring how many components are needed to describe the data) and the other is dataset fractal analysis (estimating the intrinsic dimensionality of any dataset). In both cases, the maximum number for such a plain visuo-motor task seems to be around 50.
6. The number 50 is only indicative, as it is measured for specific fMRI visuo-motor experiments. In intense cognitive situations, e.g. a pilot trying to land a plane on an aircraft carrier at night with bad weather, this is probably much higher - but in he same order of magnitude. On the other hand, when very small activations are ruled out (pre-processing by voxel smoothing), this number becomes much lower.
7. Currently, we have no idea how to develop a fully functional "brain" just by putting together 10 or 50 or even 1000 parallel processes. The simple idea of the data-driven approach is to point out that we should focus on independent -neural networks- rather than -single neurons- when trying to simulate an actual brain.
8. The current state-of-the-art neuromorphic chip by IBM provides just about 1/3 of a single voxel with 1/40 of neuron synapses within, so it is imparative to see how we can use these resources the best we can.
I hope these hints make things a bit clearer now:-)
While doing my final year of PhD, short draft Matlab routine for classifier combination...On the plastic pad we use in scuba diving for marking air consumption, nav turns and stuff. A moment of revelation, a few days of blissful coding and running experiments next.
From smaller to larger data volumes, humans tend to employ:
20 items: selection (direct) sort 20-50 items: something like bubblesort or multiple splits, e.g. quicksort-like 50-500 items: definately heapsort, mergesort (without knowing it) 500+ items: just give up, or do it as a hobby over the holidays...
For anyone that has missed what is going on in Greece since 2010, with the extreme austerity packages and the humanitarian crisis that is evolving because of them for 4 years now, this decision comes as a surprise. Of course, anyone living in the country knows that the upcoming elections cause similar and will cause even more extreme measures. There is no real politics going on in Greece right now, only taming the natives, by "whatever means necessary".
If your algorithm is purely arithmetic, then translate it to primitves (+,-,*,/) and estimate complexity based on simple full adders and flip-flps (bit level). Note that this is a very rough estimation and does not apply easilty to long, data-oriented code, since in that case your interest is with the data storage, not the operations on them (imagine adding +1 to a billion-billion-cell vector of counters).
If your algorithm is mixed-form, then you must know your hardware capabilities and, preferrably, its firmware. if you can transform your flowchart (low level) design to assembly code, then you can lookup the necessary opcodes in some standard IC and estimate (again, roughly) the order of the required IC in your case. For example, if you want to sort some 100-element vector of 16-bit integers, then a few generic x86 opcodes are enough - therefore, even 8086 (or a fraction of it) will do the job.
Generally speaking, the algorithm-to-gates estimation works only on very primitive or streamlined procedures, mostly arithmetic. That is, only when we are speaking about DSP (not CPU) implementations, like for GPUs. In almost any other case, most IC circuitry comes from the corresponding memory/heap modules, I/O, registers, etc, as the "algorithm" will require much more than a simple processing unit.
If your company knows what it's doing on all aspects (project management - SPM, quality assurance - SQA, software developing - SE), work schedules should be multi-layered and adaptable - per workpackage, per department, even per developer (special skills). Then, whenever fallbacks happen, micro-management adjustments should cover any issues, provided that they already took that in mind during the initial project plan (not very tight workflows, strong support on critical paths, provision for people's slack/sick time, etc).
If your company is not experienced in SQA and/or SPM, but has people with strong software engineering skills, then they should probably switch to agile methodologies, e.g. Scrum. This requires minimal to no SPM, moderate SQA (formal procedures) and only 2-level "loose" planning, one for 30-day sprints (e.g. next package update for release) and one daily/weekly schedule that gets updated every morning (short runs, constant rechecking). Agile can be problematic when dealing with strict deadlines, strict client contracts, strict budgets, etc, but it is worth it when the company really has high trust on its people and their work.
If your company has no expertise in neither SQA, SPM or SE, then start from the last. Good software can be produced out-of-deadline; good deadline management does not guarantee anything about the final product itself.
Slashdot Mirror
Anyone with proper Win7/Win8.x license will probably be able to sue M$ for this. Since their technical support is still active and the hardware at setup time was accepted as compatible, this is a violation of EULA. It is like buying a car and then the company suddenly changes the left-side steering wheel with a right-side one, mandatory to keep its service active. Typical M$. Switch to Linux now.
Great. Can now Nestle do the same with cocoa, so that children in Africa do not have to work 16 hours daily for $2 in order for the company to make the chocolate? Google's AI promos are plain stupid, but this is gross. This science-for-marketing thing has gone too far.
The petition has already been proven a joke. It was hijacked by bots all over the globe. I am surprised that the UK gov actually made a statement about this. Well, maybe not surprised at all...
Simple math:
Earth circumference is about 40075 km associated with 360 degrees of angle (lat,lon). This means at most 111.32 km per degree, i.e., less than 31 m per angular sec. If we use GPS coords and two typical 'float' (32-bit) with epsilon accuracy at least 10e-5 we can have 5-digit decimals or roughly 1.2 m resolution, which is far smaller than the 5-8 m maximum resolution of commercial GPS.
So, with just 8 bytes we can have more than enough location registration anywhere on the Earth with the maximum available technology today. Unless these guys have found a way to pack 3 readable words in just 8 chars with no collisions whatsoever, they deserve the next Turing award - or the "bubble of the year" award in any other case.
The local Afghan governor said yesterday that "sadly, they had to to di", because Talibans were passing through, getting treatment from MSF and borrowing (temporary stealing) the cars. Also, there is no other airforce in the area and no other hospital in that region, located ***between*** two Taliban positions (see map).
It was a clear-cut tactical decision to destroy the hospital and they kept on bombing it for almost a full hour until the central building was leveled to the ground. The MSF central has already announced that they are pulling out of Kunduz completely, which was in fact the real goal behind the bombing. If they go back there, they will be bombed again, and again, and again...
The author has already given full-access licence with the latest version of the software, non-revokable, back in 2011. If an update is scheduled, only THEN he can alter the distribution and usage selectively.
Furthermore, there is no legal way to back this up, since the international trieaties forbid any customer discrimination based on race, religion, political views, etc. This means that someone WILL use his software and if he decides to prosecute him/her, the court will drop the charges.
In practice, there are several thousands of scientist in other countries that will be willing to help. Send them your scripts and they will run them for you. No one can bring racism into science. Period.
Seriously, this is perhaps the most ridiculous "study" ever from OECD. It is nothing less than blaming the use of pens instead of pencils for bad exam grades.
Someone should eventually get in the media and apologize for this as a hoax. And do it fast.
1. Yes, the brain is massively parallel and "analog" - BUT not every neuron forms a distinct cognitive function (FBNs) and neuron do fire in pulses/spikes (almost binary) rather than continuous (analog) outputs.
2. No, the article does NOT identify 'cpu cores' in the brain. It uses this metaphor (stated clearly in the paper as such) to point out the level of parallelism needed to run anything remotely similar to the complete functional 'package' in the brain.
3. The resolution of modern fMRI is at 3mm^3 (30K-50K active voxels) but this has to do more with the localization of the activations and less with the inherent complexity (dimensionality) of the spanned data space. In other words, in this work e.g. the visual center is detected as activated or not, regardless of how fine the resolution is.
4. The fMRI captures the complete 3-D brain volume, hence the detected activations include all the "always on" circuitry like respiration, cardiac rhythm, etc. Cognitive processes are only a few of these activations and are identified by experts when looking at the actual activation maps.
5. The methodology is completely data-driven and it includes two very popular non-parametric approaches: one is ICA for blind-source separation (measuring how many components are needed to describe the data) and the other is dataset fractal analysis (estimating the intrinsic dimensionality of any dataset). In both cases, the maximum number for such a plain visuo-motor task seems to be around 50.
6. The number 50 is only indicative, as it is measured for specific fMRI visuo-motor experiments. In intense cognitive situations, e.g. a pilot trying to land a plane on an aircraft carrier at night with bad weather, this is probably much higher - but in he same order of magnitude. On the other hand, when very small activations are ruled out (pre-processing by voxel smoothing), this number becomes much lower.
7. Currently, we have no idea how to develop a fully functional "brain" just by putting together 10 or 50 or even 1000 parallel processes. The simple idea of the data-driven approach is to point out that we should focus on independent -neural networks- rather than -single neurons- when trying to simulate an actual brain.
8. The current state-of-the-art neuromorphic chip by IBM provides just about 1/3 of a single voxel with 1/40 of neuron synapses within, so it is imparative to see how we can use these resources the best we can.
I hope these hints make things a bit clearer now :-)
Even in this, he tried to mimic the big rival (Steve Jobs).
The guy never learns.
While doing my final year of PhD, short draft Matlab routine for classifier combination ...On the plastic pad we use in scuba diving for marking air consumption, nav turns and stuff.
A moment of revelation, a few days of blissful coding and running experiments next.
From smaller to larger data volumes, humans tend to employ:
20 items: selection (direct) sort
20-50 items: something like bubblesort or multiple splits, e.g. quicksort-like
50-500 items: definately heapsort, mergesort (without knowing it)
500+ items: just give up, or do it as a hobby over the holidays...
For anyone that has missed what is going on in Greece since 2010, with the extreme austerity packages and the humanitarian crisis that is evolving because of them for 4 years now, this decision comes as a surprise.
Of course, anyone living in the country knows that the upcoming elections cause similar and will cause even more extreme measures. There is no real politics going on in Greece right now, only taming the natives, by "whatever means necessary".
If your algorithm is purely arithmetic, then translate it to primitves (+,-,*,/) and estimate complexity based on simple full adders and flip-flps (bit level). Note that this is a very rough estimation and does not apply easilty to long, data-oriented code, since in that case your interest is with the data storage, not the operations on them (imagine adding +1 to a billion-billion-cell vector of counters).
If your algorithm is mixed-form, then you must know your hardware capabilities and, preferrably, its firmware. if you can transform your flowchart (low level) design to assembly code, then you can lookup the necessary opcodes in some standard IC and estimate (again, roughly) the order of the required IC in your case. For example, if you want to sort some 100-element vector of 16-bit integers, then a few generic x86 opcodes are enough - therefore, even 8086 (or a fraction of it) will do the job.
Generally speaking, the algorithm-to-gates estimation works only on very primitive or streamlined procedures, mostly arithmetic. That is, only when we are speaking about DSP (not CPU) implementations, like for GPUs. In almost any other case, most IC circuitry comes from the corresponding memory/heap modules, I/O, registers, etc, as the "algorithm" will require much more than a simple processing unit.
If your company knows what it's doing on all aspects (project management - SPM, quality assurance - SQA, software developing - SE), work schedules should be multi-layered and adaptable - per workpackage, per department, even per developer (special skills). Then, whenever fallbacks happen, micro-management adjustments should cover any issues, provided that they already took that in mind during the initial project plan (not very tight workflows, strong support on critical paths, provision for people's slack/sick time, etc).
If your company is not experienced in SQA and/or SPM, but has people with strong software engineering skills, then they should probably switch to agile methodologies, e.g. Scrum. This requires minimal to no SPM, moderate SQA (formal procedures) and only 2-level "loose" planning, one for 30-day sprints (e.g. next package update for release) and one daily/weekly schedule that gets updated every morning (short runs, constant rechecking). Agile can be problematic when dealing with strict deadlines, strict client contracts, strict budgets, etc, but it is worth it when the company really has high trust on its people and their work.
If your company has no expertise in neither SQA, SPM or SE, then start from the last. Good software can be produced out-of-deadline; good deadline management does not guarantee anything about the final product itself.