Will read that, thank you. Certainly, it is true that some information is better than none. I just expected (naively perhaps) that something called an 'index' would be normalized to some attribute like population (or one of the others you named).
Clarification: The phone was not an iPhone 4s, but an iPhone 6s. The "Die Welt" article only quotes a policeman as saying that "iPhones from 4s upwards cannot be accessed without the PIN", but makes no further statement as to the actual phone used by the accused. The original source, however, clearly states the phone model as an iPhone 6s:
Thank you for one of the few comments in this thread that actually deals with what this is (as opposed to what it isn't, i.e. human-level AI).
I would like to add that hyperparameter tuning is _not_ a trivial part of programming a machine learning model, therefore this IS something rather interesting. It lowers the effort needed to do something interesting with machine learning, and therefore makes machine learning much more accessible to non-experts.
However, the tasks in machine learning that still require humans need a much more flexible sort of intelligence: a) asking the right question, i.e. determining the variables to be predicted and b) finding the right input parameters (or independent variables) that will help answer the question.
And 4.7% is just a lower bound for the real number because they did not check those who never took part in a drug test. The share of the total population is probably closer to 10%.
With these sorts of numbers, it's just ludicrous to keep up prohibition.
I hope this will also lead to improvements in consumer devices for people without hearing impairment. I am always jealous when I see what an elegant device my father's hearing aid is. All the Bluetooth headsets that I have tried, even at the very high end, are extremely clumsy beasts in comparison. If the experience from lower-end hearing aids trickles down to high-end Bluetooth headsets, that will improve my life considerably.
This is not a question of wanting to do something. Germany (like most other countries) has signed the 1951 Refugee Convention and the 1967 Protocol. This means that we cannot legally refuse helping refugees that arrive here. Neither can Austria, the US and 143 other countries. If we now refuse to honor that convention, it will for all practical purposes cease to exist.
Furthermore, I second that we do not really have a problem with the number of refugees in Germany. Temporary issues, yes, but in the long term this will not be a problem.
So basically, Facebook will provide a command-line interface to lots of services. I will not use it because Facebook, but I like the direction this is heading.
You are spot on, though I consider (offline) assistance systems to be less critical than you do. After all, few seemingly mechanical functions of cars are now purely mechanical. With the weight of current cars, it would be very tiresome e.g. to live without servo steering.
Is it just me, or should cars just not come with (much) wireless connectivity?
Granted, a key that opens doors remotely is useful. And it is nice to have a Bluetooth connection to the car radio. But apart from that?
Having the car online most of the time (which is what OTA updates would necessitate) has more downside than upside. In such a life-critical system, the additional convenience of an Internet connection does not justify the reduced safety. I personally never want to see a "Transfer x bitcoins to address y now, and you will get your steering back" message on my dashboard.
"If I were a younger man, I would write a history of human stupidity; and I
would climb to the top of Mount McCabe and lie down on my back with my history
for a pillow; and I would take from the ground some of the blue-white poison that
makes statues of men; and I would make a statue of myself, lying on my back,
grinning horribly, and thumbing my nose at You Know Who." --Bokonon
I used the term "significant" rather in the statistical sense. As in, "Is there a high probability that the reduction in cancer incidence is based on the vitamin D and not on random variation in the sample". This probability can be calculated, but only based on the original data, not on the cumulated data (60%) given in the article. A typical probability that would validate the results (in medical science, not in sciences that benefit from better repeatability of experiments) is 95%.
This is one of the principal problems of medical science which makes it nearly impossible to prove the effectiveness of a therapy.
Age-adjusted cancer incidence in females in 2003 was 0.4036%. In favour of the article's publishers, I assume there was no control group to reduce the numbers. ~4.8 women on average would have gotten cancer in one year. The study was done over a time of four years, so it would have been 20. Now reduce that by 60% and you arrive at 8 women with cancer, or a difference of 12.
We see that we have probably been fooled by the percentages. A lower incidence of 60% sounds high, but 12 cases less?
Checking the significance level isn't really possible unless you have the complete setup and data of the study (statisticians, please correct me if I'm wrong!). However, I suspect that the answer in a detailed analysis would have to be "maybe". I guess I'll have to wait for the final publication of the results until I begin to sunbathe during lunch break.
Slashdot Mirror
Will read that, thank you. Certainly, it is true that some information is better than none. I just expected (naively perhaps) that something called an 'index' would be normalized to some attribute like population (or one of the others you named).
True. Thanks for pointing that out. I wish I had moderator points to upvote your post.
I never thought it possible Statista would publish such a sloppily defined metric.
This makes me pity the people in Iraq even more...
You should check your figures.
Global Terrorism Index 2018: https://www.statista.com/stati...
The US have a higher prevalence of terrorism than any European country (although Ukraine is only one place behind the US).
Clarification: The phone was not an iPhone 4s, but an iPhone 6s. The "Die Welt" article only quotes a policeman as saying that "iPhones from 4s upwards cannot be accessed without the PIN", but makes no further statement as to the actual phone used by the accused. The original source, however, clearly states the phone model as an iPhone 6s:
http://www.badische-zeitung.de...
Thank you for one of the few comments in this thread that actually deals with what this is (as opposed to what it isn't, i.e. human-level AI).
I would like to add that hyperparameter tuning is _not_ a trivial part of programming a machine learning model, therefore this IS something rather interesting. It lowers the effort needed to do something interesting with machine learning, and therefore makes machine learning much more accessible to non-experts.
However, the tasks in machine learning that still require humans need a much more flexible sort of intelligence: a) asking the right question, i.e. determining the variables to be predicted and b) finding the right input parameters (or independent variables) that will help answer the question.
...Normal.
And 4.7% is just a lower bound for the real number because they did not check those who never took part in a drug test. The share of the total population is probably closer to 10%.
With these sorts of numbers, it's just ludicrous to keep up prohibition.
Wait... Shouldn't that have been "Vast Autonomic Logistics Information System"?
I hope this will also lead to improvements in consumer devices for people without hearing impairment. I am always jealous when I see what an elegant device my father's hearing aid is. All the Bluetooth headsets that I have tried, even at the very high end, are extremely clumsy beasts in comparison. If the experience from lower-end hearing aids trickles down to high-end Bluetooth headsets, that will improve my life considerably.
This is not a question of wanting to do something. Germany (like most other countries) has signed the 1951 Refugee Convention and the 1967 Protocol. This means that we cannot legally refuse helping refugees that arrive here. Neither can Austria, the US and 143 other countries. If we now refuse to honor that convention, it will for all practical purposes cease to exist.
Furthermore, I second that we do not really have a problem with the number of refugees in Germany. Temporary issues, yes, but in the long term this will not be a problem.
I know this thread is going to be full of We Hate Americans - it's already started.
The overwhelming feeling I get is not hate, but sympathy.
So basically, Facebook will provide a command-line interface to lots of services. I will not use it because Facebook, but I like the direction this is heading.
You are spot on, though I consider (offline) assistance systems to be less critical than you do. After all, few seemingly mechanical functions of cars are now purely mechanical. With the weight of current cars, it would be very tiresome e.g. to live without servo steering.
Is it just me, or should cars just not come with (much) wireless connectivity?
Granted, a key that opens doors remotely is useful. And it is nice to have a Bluetooth connection to the car radio. But apart from that?
Having the car online most of the time (which is what OTA updates would necessitate) has more downside than upside. In such a life-critical system, the additional convenience of an Internet connection does not justify the reduced safety. I personally never want to see a "Transfer x bitcoins to address y now, and you will get your steering back" message on my dashboard.
"If I were a younger man, I would write a history of human stupidity; and I would climb to the top of Mount McCabe and lie down on my back with my history for a pillow; and I would take from the ground some of the blue-white poison that makes statues of men; and I would make a statue of myself, lying on my back, grinning horribly, and thumbing my nose at You Know Who." --Bokonon
I used the term "significant" rather in the statistical sense. As in, "Is there a high probability that the reduction in cancer incidence is based on the vitamin D and not on random variation in the sample". This probability can be calculated, but only based on the original data, not on the cumulated data (60%) given in the article. A typical probability that would validate the results (in medical science, not in sciences that benefit from better repeatability of experiments) is 95%.
This is one of the principal problems of medical science which makes it nearly impossible to prove the effectiveness of a therapy.
Science... Medical science... statistics... a million alarm bells go off in my head.
With the limited data that the article gives, I tried to do the math.
US cancer incidence statistics for 2003 are available here: http://www.cdc.gov/cancer/npcr/npcrpdfs/US_Cancer_ Statistics_2003_Incidence_and_Mortality.pdf (PDF)
Age-adjusted cancer incidence in females in 2003 was 0.4036%. In favour of the article's publishers, I assume there was no control group to reduce the numbers. ~4.8 women on average would have gotten cancer in one year. The study was done over a time of four years, so it would have been 20. Now reduce that by 60% and you arrive at 8 women with cancer, or a difference of 12.
We see that we have probably been fooled by the percentages. A lower incidence of 60% sounds high, but 12 cases less?
Checking the significance level isn't really possible unless you have the complete setup and data of the study (statisticians, please correct me if I'm wrong!). However, I suspect that the answer in a detailed analysis would have to be "maybe". I guess I'll have to wait for the final publication of the results until I begin to sunbathe during lunch break.