Results matter more than a 'peers' opinion of the results.
You misunderstand the process. Peer opinions are based on the results. They are also based on years of study leading to an appreciation of what results are actually 1) interesting and 2) useful. These are crude words for the distinction, but to illustrate, if AlphaFold were to work perfectly it would only be useful. It wouldn't improve understanding and thereby advance science beyond making some specific current task potentially easier. (Even if it might be really great for engineering).
If the training set contains all the magic rules
There's good reason to think this training set doesn't contain all the magic rules. What the AlphaFold team should do is use structures solved before 2005 to train their model, and structures with novel folds solved after 2005 to test. If they can achieve very high absolute performance in that context, all critics will be silenced.
I wonder how much the network has been trained to recognize existing (evolutionary dependent) protein families and their patterns vs. a new random sequence folder.
That's why they should use the historical validation approach! Train on structures solved before 2005, then predict only novel folds solved after 2005. Perform well in that context and I'll be impressed.
The former may be just as useful in practice but may teach us a bit less about the mechanics of folding.
Unlike the physics-based and statistical potential methods, can the DeepMind approach ever contribute to understanding how proteins fold? IMHO that's an open question, and one that's critical to their presumably forthcoming publication. For example, do their features weights say something interesting about cation-pi interactions? Rosetta infamously ignores cation-pi because of overfitting concerns (even though cation-pi can be very structurally important).
Google's team of 10 people produced a better result with 2 years of work than the entire academic field has been able to produce in the last 30
That's not a correct reading of the results. First, previous efforts are based on putative understanding about how proteins fold. Obviously, this understanding is incomplete - or the physics based methods would perform better. (Even statistical potentials like in Rosetta are physics based in important ways). Second, DeepMind isn't even on the radar in the server component of CASP. The server competition is intrinsically more difficult because it requires robust software that isn't highly dependent on user parameters. Rosetta for example is ~20th in the general competition and 4th place in the server competition.
Finally, DeepMind has not demonstrated the historical performance of their approach. They should see how well novel protein folds solved after e.g. 2005 are predicted using only structures solved before 2005 to train. To the extent that Rosetta works, it works in such an environment. In fact, one its first results was a novel fold (Top7).
That kind of road structure is not unusual in Netherland.
Based on my one trip to the Netherlands, I'd guess there are also bike / foot paths that cut through to the main road. In US suburbs there are usually no such paths, so the main road is essentially inaccessible without a car.
As an aside, the current PC terms are "developmentally disabled" and "developmentally delayed," which have the benefit of being more correct/explanatory descriptors with an important an obvious distinction. (Though just as easily twisted into insults).
Tech basically doesn't matter in the long term housing cost trend for the San Francisco area. The rate of increase has been unchanged for many decades, laying the blame squarely at the feet of development-killing regulation/NIMBYs.
We can't get GPUs for biology, either. Molecular dynamics simulations, protein structure determination using cryo-EM, and a number of other areas in computational biology rely on GPU-accelerated computing for obtaining results in a timely fashion.
GPU computing in this area has the particular advantage over CPU clusters in rapid iteration over user-defined parameters. If you know a correct set of parameters, running a job for several days on CPU is fine. If you need to run 100 such jobs to find those parameters, it's totally inadequate.
It's very frustrating that the cost of a 4-GPU workstation has gone from ~$6,000 to ~$12,000 in the past year. It especially hinders smaller labs and institutions. Ethereum can't crash soon enough (bitcoin is mined on ASICS, Ethereum is the real driver of GPU price now).
Or Facebook App just reads your gchat data directly.
Maybe not that crazy since most phones have Facebook as a system app and they've already been outed for using the microphone to bug people for ad targeting.
Sure, but this shows where many of them are all at once, for free, with GPS coordinates. People run on the roads mostly, so it also gives a road map of the base. Go zoom in on some of those random hotspots in podunk Afghanistan, it's pretty weird.
This is a non sequitur. Sure, this store might have issues that limit its ability to replace cashiers at the present time ("ability," because elderly people are a significant market segment that would be unprofitable to ignore). For the moment, so do self-driving cars (*cough* inclement weather *cough*).
The point is that there's something fundamentally wrong with a system wherein elimination of menial labor is somehow bad.
You can't have an economy based on consumption when no one can afford to buy because they don't have jobs to earn the money that it takes to consume.
Exactly. Also, both basic income and jobs guarantee could virtually eliminate a host of other social ills, like homelessness and food insecurity. Those things directly make all of our lives worse, e.g. by forcing us to walk through feces encrusted encampments and increasing the incidence of property crimes.
Further, unlike top-heavy tax cuts, bottom-heavy cash transfers for folks with marginal propensity to consume near 1 will drive growth and thereby limit their impact on net government revenue.
Unemployment insurance is just a holdover from another era, dressed up to appeal to old boomers too lazy to work and too entitled to seek training or personal betterment.
Instead of paying people NOT to work - the big government idea favored by the AC - we should allow federal agencies (especially parks and transportation) to hire unlimited minimum wage workers for infrastructure improvement projects or paid training. This approach eliminates other wage regulation (the private sector must pay higher than the guarantee wage), delivers the ultimate work requirement for government assistance, and provides a direct avenue to labor force retraining/modernization. It's far simpler than the current system involving complex, overlapping big government programs, more economically useful (infrastructure building and maintenance), and more socially useful for able-bodied people (training opportunities, work requirements, etc.).
A jobs guarantee is THE conservative answer to welfare, and it's a shame you (the AC) are too close-minded to see it.
The political economy is broken: innovation that delivers broad productivity and standard-of-living increases is "bad" because it puts people out of work. This phenomenon is not new. For example, some metro transit systems rolled out in the 70's were designed for total automation, but were forced to employ operators by unions and/or public outcry.
There are two simple, direct fixes that should be on the table. One is a basic income, the other a jobs guarantee.
Fortunately, we haven’t found password theft on the 50,000 sites that we analyzed. Instead, we found tracking scripts embedded by the first party abusing the same technique to extract emails addresses for building tracking identifiers.
Why would the first party need to steal my email address / username? I just used it to log in to their site!
Indeed. It however does work when you manually tell LastPass to fill a password. But nothing in the UI prompts you to do so, so I consider this safe behavior.
The important thing is what happens if there is a real login field and the third party fields are hidden.
It looks like the only safe way to use LastPass is copy-and-paste. Which, in retrospect, makes sense.
Certain genres, at least, are more like "real life" than e.g. image classification and other tasks commonly approached using machine learning. A program that can play Starcraft, or GTA, or Grand Turismo, or Counterstrike using the same inputs and outputs as a human player is a lot closer to programs for real-world tasks like driving a car, than a program that identifies handwritten characters or plays Go.
I agree completely, but I do think that a chemical - i.e. a drug - of some kind was very likely used against these people. There are absolutely drugs that can make you hear things and develop SIB (swelling, itching brain).
I'm typing this on a 2015 MBP, given to me by my employer. It definitely has some things to recommend it, e.g. it's light weight, decent battery life, easy access to *nix tools (via Homebrew), speaker capability and screen brightness. In other respects though, I have to agree with the submitter. Hardware-wise, it's about on par with my 2010 Thinkpad. OS-wise there are a bunch of deficiencies which are not just my opinion about look-and-feel, but actual missing features. I'll just describe one quickly, which I feel is emblematic of Apple's general issues.
On a Mac, you can switch through display elements (windows, dialogs, full screen apps) in two segregated ways. Cmd-tab switches applications, cmd-backtick switches windows within an application. On one level, the segregation is logical, but in practice it leads to some really inelegant behaviors. It's impossible to place one window on top of a fullscreen application, so among other things you can't take notes while watching a fullscreen video. Full screen applications create their own workspaces which are children of the original workspace, and switching back to other workspaces isn't allowed. Actually, you can switch, but it will immediately scroll back to the full screen application.
Windows, on the other hand, simply has alt-tab (or win-tab), which cycles through all display elements without regard for parent application. It naturally allows windows to be displayed above fullscreen applications, and for fullscreen applications to be left in fullscreen mode when switching away or minimizing. It's more simplistic, but also more functional. Again, that's not an opinion, it's a missing feature: on a Windows PC one can take notes on a fullscreen video, and on a Mac one cannot.
It's a basic design choice that seems logical and elegant, but ends up handicapping the window system down the line. Another similar example is the total lack of a hotkey to restore minimized windows. There is Hide (cmd-H), but it only works on entire applications at a time.
Slashdot Mirror
Results matter more than a 'peers' opinion of the results.
You misunderstand the process. Peer opinions are based on the results. They are also based on years of study leading to an appreciation of what results are actually 1) interesting and 2) useful. These are crude words for the distinction, but to illustrate, if AlphaFold were to work perfectly it would only be useful. It wouldn't improve understanding and thereby advance science beyond making some specific current task potentially easier. (Even if it might be really great for engineering).
If the training set contains all the magic rules
There's good reason to think this training set doesn't contain all the magic rules. What the AlphaFold team should do is use structures solved before 2005 to train their model, and structures with novel folds solved after 2005 to test. If they can achieve very high absolute performance in that context, all critics will be silenced.
I wonder how much the network has been trained to recognize existing (evolutionary dependent) protein families and their patterns vs. a new random sequence folder.
That's why they should use the historical validation approach! Train on structures solved before 2005, then predict only novel folds solved after 2005. Perform well in that context and I'll be impressed.
The former may be just as useful in practice but may teach us a bit less about the mechanics of folding.
Unlike the physics-based and statistical potential methods, can the DeepMind approach ever contribute to understanding how proteins fold? IMHO that's an open question, and one that's critical to their presumably forthcoming publication. For example, do their features weights say something interesting about cation-pi interactions? Rosetta infamously ignores cation-pi because of overfitting concerns (even though cation-pi can be very structurally important).
Google's team of 10 people produced a better result with 2 years of work than the entire academic field has been able to produce in the last 30
That's not a correct reading of the results. First, previous efforts are based on putative understanding about how proteins fold. Obviously, this understanding is incomplete - or the physics based methods would perform better. (Even statistical potentials like in Rosetta are physics based in important ways). Second, DeepMind isn't even on the radar in the server component of CASP. The server competition is intrinsically more difficult because it requires robust software that isn't highly dependent on user parameters. Rosetta for example is ~20th in the general competition and 4th place in the server competition.
Finally, DeepMind has not demonstrated the historical performance of their approach. They should see how well novel protein folds solved after e.g. 2005 are predicted using only structures solved before 2005 to train. To the extent that Rosetta works, it works in such an environment. In fact, one its first results was a novel fold (Top7).
Who believes Ajit Pai actually cares about or is working on this?
Random name-calling, totally orthogonal to the comment thread, is an interesting use of your time.
That kind of road structure is not unusual in Netherland.
Based on my one trip to the Netherlands, I'd guess there are also bike / foot paths that cut through to the main road. In US suburbs there are usually no such paths, so the main road is essentially inaccessible without a car.
Interesting that your sig is labeling anyone who uses the term SJW a fuckwit
No, it's labelling anyone who uses the term SJW a fuckwit, after correctly identifying it as a synonymous with "I don't like you."
As an aside, the current PC terms are "developmentally disabled" and "developmentally delayed," which have the benefit of being more correct/explanatory descriptors with an important an obvious distinction. (Though just as easily twisted into insults).
Tech basically doesn't matter in the long term housing cost trend for the San Francisco area. The rate of increase has been unchanged for many decades, laying the blame squarely at the feet of development-killing regulation/NIMBYs.
We can't get GPUs for biology, either. Molecular dynamics simulations, protein structure determination using cryo-EM, and a number of other areas in computational biology rely on GPU-accelerated computing for obtaining results in a timely fashion.
GPU computing in this area has the particular advantage over CPU clusters in rapid iteration over user-defined parameters. If you know a correct set of parameters, running a job for several days on CPU is fine. If you need to run 100 such jobs to find those parameters, it's totally inadequate.
It's very frustrating that the cost of a 4-GPU workstation has gone from ~$6,000 to ~$12,000 in the past year. It especially hinders smaller labs and institutions. Ethereum can't crash soon enough (bitcoin is mined on ASICS, Ethereum is the real driver of GPU price now).
collusion
Or Facebook App just reads your gchat data directly.
Maybe not that crazy since most phones have Facebook as a system app and they've already been outed for using the microphone to bug people for ad targeting.
iOS 11 assistive typing, I think. In some other threads people have been spamming the instructions to turn it off.
Sure, but this shows where many of them are all at once, for free, with GPS coordinates. People run on the roads mostly, so it also gives a road map of the base. Go zoom in on some of those random hotspots in podunk Afghanistan, it's pretty weird.
Look up the difference between labor force participation, and unemployment.
This is a non sequitur. Sure, this store might have issues that limit its ability to replace cashiers at the present time ("ability," because elderly people are a significant market segment that would be unprofitable to ignore). For the moment, so do self-driving cars (*cough* inclement weather *cough*).
The point is that there's something fundamentally wrong with a system wherein elimination of menial labor is somehow bad.
You can't have an economy based on consumption when no one can afford to buy because they don't have jobs to earn the money that it takes to consume.
Exactly. Also, both basic income and jobs guarantee could virtually eliminate a host of other social ills, like homelessness and food insecurity. Those things directly make all of our lives worse, e.g. by forcing us to walk through feces encrusted encampments and increasing the incidence of property crimes.
Further, unlike top-heavy tax cuts, bottom-heavy cash transfers for folks with marginal propensity to consume near 1 will drive growth and thereby limit their impact on net government revenue.
Unemployment insurance is just a holdover from another era, dressed up to appeal to old boomers too lazy to work and too entitled to seek training or personal betterment.
Instead of paying people NOT to work - the big government idea favored by the AC - we should allow federal agencies (especially parks and transportation) to hire unlimited minimum wage workers for infrastructure improvement projects or paid training. This approach eliminates other wage regulation (the private sector must pay higher than the guarantee wage), delivers the ultimate work requirement for government assistance, and provides a direct avenue to labor force retraining/modernization. It's far simpler than the current system involving complex, overlapping big government programs, more economically useful (infrastructure building and maintenance), and more socially useful for able-bodied people (training opportunities, work requirements, etc.).
A jobs guarantee is THE conservative answer to welfare, and it's a shame you (the AC) are too close-minded to see it.
The political economy is broken: innovation that delivers broad productivity and standard-of-living increases is "bad" because it puts people out of work. This phenomenon is not new. For example, some metro transit systems rolled out in the 70's were designed for total automation, but were forced to employ operators by unions and/or public outcry.
There are two simple, direct fixes that should be on the table. One is a basic income, the other a jobs guarantee.
I guess it only says "embedded by" so they may not be aware, or may be complicit in allowing the third-party to also get this information.
Fortunately, we haven’t found password theft on the 50,000 sites that we analyzed. Instead, we found tracking scripts embedded by the first party abusing the same technique to extract emails addresses for building tracking identifiers.
Why would the first party need to steal my email address / username? I just used it to log in to their site!
Indeed. It however does work when you manually tell LastPass to fill a password. But nothing in the UI prompts you to do so, so I consider this safe behavior.
The important thing is what happens if there is a real login field and the third party fields are hidden.
It looks like the only safe way to use LastPass is copy-and-paste. Which, in retrospect, makes sense.
They had to take care of some outstanding bugs and feature requests for the death star attack run simulator our astronauts use for training.
Certain genres, at least, are more like "real life" than e.g. image classification and other tasks commonly approached using machine learning. A program that can play Starcraft, or GTA, or Grand Turismo, or Counterstrike using the same inputs and outputs as a human player is a lot closer to programs for real-world tasks like driving a car, than a program that identifies handwritten characters or plays Go.
I agree completely, but I do think that a chemical - i.e. a drug - of some kind was very likely used against these people. There are absolutely drugs that can make you hear things and develop SIB (swelling, itching brain).
I'm typing this on a 2015 MBP, given to me by my employer. It definitely has some things to recommend it, e.g. it's light weight, decent battery life, easy access to *nix tools (via Homebrew), speaker capability and screen brightness. In other respects though, I have to agree with the submitter. Hardware-wise, it's about on par with my 2010 Thinkpad. OS-wise there are a bunch of deficiencies which are not just my opinion about look-and-feel, but actual missing features. I'll just describe one quickly, which I feel is emblematic of Apple's general issues.
On a Mac, you can switch through display elements (windows, dialogs, full screen apps) in two segregated ways. Cmd-tab switches applications, cmd-backtick switches windows within an application. On one level, the segregation is logical, but in practice it leads to some really inelegant behaviors. It's impossible to place one window on top of a fullscreen application, so among other things you can't take notes while watching a fullscreen video. Full screen applications create their own workspaces which are children of the original workspace, and switching back to other workspaces isn't allowed. Actually, you can switch, but it will immediately scroll back to the full screen application.
Windows, on the other hand, simply has alt-tab (or win-tab), which cycles through all display elements without regard for parent application. It naturally allows windows to be displayed above fullscreen applications, and for fullscreen applications to be left in fullscreen mode when switching away or minimizing. It's more simplistic, but also more functional. Again, that's not an opinion, it's a missing feature: on a Windows PC one can take notes on a fullscreen video, and on a Mac one cannot.
It's a basic design choice that seems logical and elegant, but ends up handicapping the window system down the line. Another similar example is the total lack of a hotkey to restore minimized windows. There is Hide (cmd-H), but it only works on entire applications at a time.