Skip to content

Author: Briggs

January 11, 2018 | 4 Comments

Facial Recognition Makes Mistakes. No Big Deal, Right?

Stream: Should Big Brother Watch Us If He Would Keep Us Safe?

China like Britain is installing cameras in public places to track its citizens. Britain already has at least 1 surveillance camera for every 11 people, a fraction that is rising. China wants in on the photographic fun. The Washington Post reports:

The intent is to connect the security cameras that already scan roads, shopping malls and transport hubs with private cameras on compounds and buildings, and integrate them into one nationwide surveillance and data-sharing platform.

It will use facial recognition and artificial intelligence to analyze and understand the mountain of incoming video evidence; to track suspects, spot suspicious behaviors and even predict crime; to coordinate the work of emergency services; and to monitor the comings and goings of the country’s 1.4 billion people, official documents and security industry reports show.

Computers Make Mistakes

“Artificial intelligence” (a.k.a. “deep learning”) always sounds scary, but it is nothing more than old-fashioned statistical modeling done on a large scale. Because it is a form of modeling, it is imperfect. That means that when an algorithm designed to look at a picture of Mr. Wong and say, “This is Mr. Wong”, sometimes it won’t. Sometimes it will say it is you.

What harm could there be in that?

Consider that you have been incorrectly identified as standing outside a certain building where known troublemakers have been seen. The algorithm that said you were there then looks to the “Police Cloud” database that has “such data as criminal and medical records, travel bookings, online purchase and even social media comments.”

The computer next looks up the “meta data” from your phone records. This tells exactly where you were when you made every call, who you called and for how long, on what device you and the other party used, whether the call was followed by any data (say, a Snapchat), and so on. The only thing the computer does not admit to knowing is what you said.

The algorithm now updates your “social credit” score, lowering it. Not only does it ding your score, but the people you called also take a small hit.

The entire process is automatic, with no way to check errors, so you’ll never know why the hiring manager rejected your application. (You won’t know at Google, either.)

We’re All Guilty

There is another possibility. The facial-recognition algorithm does not make a mistake. It really was you standing there. You may have had an innocent explanation for being at that suspicious corner. But we’re talking public safety here. Why take a chance? A suspicious corner was involved. And it’s always better to be safe than sorry, isn’t it?

Here we recall the words []

Click here to read the rest. Clear your cookies after to maintain plausible deniability.

January 10, 2018 | 18 Comments

Falsifiability Is Falsifiable

We’ve talked many times before, and at greater length in Uncertanity, about the concept of falsifiability. It has come up again lately.

The term shouldn’t be but is equivocal. I mean it is used in an equivocal sense, which is always a danger, because it happens that an argument is started with one sense of the word, but finished with another, invalidating the argument.

The first, and what I say should be the only, is the logical interpretation. To falsify is to prove something false. To prove is to demonstrate by a sound, valid argument. Thus if a man holds the proposition “7 is less than 4”, we can falsify that proposition with a simple proof which takes as its premises basic arithmetic. Indeed, for every proof we have (in math, logic, philosophy) we have also proved a contrary is false. Falsifications are thus not rare—but they are restricted, as we shall see.

Theories make predictions, or rather it is possible to infer predictions from theories. If a theory says, “X is impossible” and X is observed, then the theory has been falsified, i.e. proved false. We here take impossible in its strictest sense.

But if the theory merely says, “X is very unlikely” and X happens, then the theory has not been falsified. Not according to the logical definition of the word. Rare things will happen. The theory admitted this rare thing was not an impossibility; it admitted that the rare thing was possible. The rare thing did happen. Therefore, if anything, the theory has been given support!

Falsification was, as we all know, made popular by Karl Popper. He relied upon it for both philosophical and practical reasons. About the latter, he was frustrated by certain ridiculous theories that could not, by the use of the common tools of the day, be shown to be false. Every observation, said the believers in these queer theories, was consistent with their cherished models.

Take the theory that UFOs have visited earth. The lack of clear or unambiguous photographs means that the UFOs are clever in evading cameras. That NASA denies the existence of UFOs shows the government is in on the conspiracy. And so on. There is no way to falsify the theory that UFOs are real.

Indeed, there does not exist a definitive proof, in the logical sense. Such a proof would have to demonstrate both why other spaceship-building lifeforms could not exist or that even if they could, they could not have got to earth. These demonstrations cannot say such feats are unlikely; they have to say they are impossible. And that’s not possible. UFOs therefore cannot be falsified.

True theories also cannot be falsified; and they have the happy bonus of being right about everything. That is, every observation also supports true theories. No observation anywhere falsifies the theory “7 > 4”.

Enter the second, and incorrect, use of falsified. People will say that the UFO theory has been “practically falsified”, because no evidence of little green men has ever withstood scrutiny. These people are right about about the lack of evidence, and even right according to my lights about their judgement not to take UFOs seriously. But they are wrong to say that the theory has been falsified, practically or not.

Practically falsified is to falsified as practically a virgin is to virgin. They are not in the same philosophical ballpark.

Back to our rare X. A theory has said X is rare, but not impossible. X is seen. Depending on the rarity, for rarity is a loose word with many interpretations, people will incorrectly say the theory has been “practically falsified”. Suppose it was judged Pr(X|theory) = ε > 0, and X is observed. If ε is “small enough”, the criterion for “practical falsification” has been met. Yet once people allow themselves the phrase “practically falsified”, they generously given themselves permission to strip off the qualifier and say just plain “falsified”.

If this doesn’t sound familiar, then you have not been paying attention, dear reader. For this is the basis of the use of p-values — where ε stands in for the magic number. As some of us recall, it was Fisher’s intent to follow Popper’s lead here.

Now it may be the case that every other theory that we know also says X is rare; in notation Pr(X|other theories) = rare. Or we may not know of any other theory; or the theories we know do not allow us to compute a numerical probability.

In the logical sense, all we can say about X is that, perhaps, some theories give it higher probabilities than others. It is natural also to think that the theory which said X was more likely is itself more useful, which it might very well be. But that depends on the use of theory, and how the probabilities affect our decisions. This is why I said I agree about the decision to treat UFOs as unimportant (mistakes or hoaxes etc.), and why I did not say UFOs are false.

We are on the familiar ground of the predictive method, where we recognize that probability is not decision, and understanding how a theory can be useful to one man, but useless to a second, and so forth. See the on-going class for details.

This leads to the conclusion that “practical falsification” is an act, a decision, and not a demonstration. Nothing has been proved with “practical falsification”; something has been judged. “Practical falsification” is in this sense an opinion. In statistics, p-values are a one-size-fits-all decision, because the magic number is, well, magic. One theory is judged false (the “null”), while another has been judged true (the “alternate”).

Wait. Not judged true. Judged “not falsified”. Popper could never bring himself to believe anything; all theories were to him temporary and waiting to be supplanted. That accounts for the screwy “not falsified”, which isn’t wrong, but it is odd. See Uncertainty for more details on this.

January 9, 2018 | 20 Comments

Free Data Science Class: Predictive Case Study 1, Part VII

Review!

This is our last week of theory. Next week the practical side begins in earnest. However much fun that will be, and it will be a jolly time, this is the more important material.

Last time we learned the concept of irrelevance. A premise is irrelevant if when it is added to the model, the probability of our proposition of interest does not change. Irrelevance, like probability itself, is conditional. Here was our old example:

    (7a) Pr(CGPA = 4 | grading rules, old obs, sock color, math) = 0.05,
    (7c) Pr(CGPA = 4 | grading rules, old obs, math) = 0.05,

In the context of the premises “grading rules, old obs, math”, “sock color” was irrelevant because the probability of “CGPA = 4” did not change when adding it. It is not that sock color is unconditionally irrelevant. For instance, we might have

    (7d) Pr(CGPA = 3 | grading rules, old obs, sock color, math) = 0.10,
    (7e) Pr(CGPA = 3 | grading rules, old obs, math) = 0.12,

where now, given a different proposition of interest, sock color has become relevant. Whether it is useful is, and always will be, whether it is pertinent to any decisions we would make about CGPA = 3. We might also have:

    (7f) Pr(CGPA = 4 | grading rules, old obs, sock color) = 0.041,
    (7g) Pr(CGPA = 4 | grading rules, old obs) = 0.04,

where sock color becomes relevant to CGPA = 4 absent our math (i.e. model) assumptions. Again, all relevance is conditional. And all usefulness depends on decision.

Decision is not unrelated to knowledge about cause. Cause is not something to be had from probability models; it is something that comes before them. Failing to understand this is the cause (get it!) of confusion generated by p-values, hypothesis tests, Bayes factors, parameter estimates, and so on. Let’s return to our example:

    (7a) Pr(CGPA = 4 | grading rules, even more old obs, sock color, math) = 0.05,
    (7b) Pr(CGPA = 4 | grading rules, even old obs, math) = 0.051.

Sock color is relevant. But does sock color cause a change in CGPA? How can it? Doubtless we can think of a story. We can always think of a story. Suppose sock color indicates the presence of white or light colored socks (then, the absence of sock color from the model implies dark color or no hosiery). We might surmise light color socks reflect extra light in examination rooms, tiring the eyes of wearers so that they will be caused to miss questions slightly more frequently than their better apparelled peers.

This is a causal story. It might be true. You don’t know it isn’t. That is, you don’t know unless you understand the true cause of sock color on grades. And, for most of us, this is no causation at all. We can tell an infinite number of causal stories, all equally consistent with the calculated probabilities, in which sock color affects CGPA. There cannot be proof they are all wrong. We therefore have to use induction (see this article) to infer sock color by its nature is acausal (to grades). We must grasp the essence of socks and sock-body contacts. This is perfectly possible. But it is something we do beyond the probabilities, inferring from the particular observations to the universal truth about essence. Our comprehension of cause is not in the probabilities, nor in the observations, but in the intellectual leap we make, and must make.

This is why any attempt to harness observations to arrive at causal judgments must fail. Algorithms cannot leap into the infinite like we can. Now this is a huge subject, beyond that which we can prove in this lesson. In Uncertainty, I cover it in depth. Read the Chapter on Cause and persuade yourself of the claims made above, or accept them for the sake of argument here.

What follows is that any kind of hypothesis test (or the like) must be making some kind of error, because it is claiming to do what we know cannot be done. It is claiming to have identified a cause, or a cause-like thing, from the observations.

Now classical statistics will not usually say that “cause” has been identified, but it will always be implied. In a regression for Income on Sex, it will be claimed (say) “Men make more than women” based on a wee p-value. This implies sex causes income “gaps”. Or we might hear, if the researcher is trying to be careful, “Sex is linked to income”. “Linked to” is causal talk. I have yet to see any definition (and they are all usually long-winded) of “linked to” that did not, in the end, boil down to cause.

There is a second type of cause to consider, the friend-of-a-friend cause, or the cause of a cause (or of a cause etc.). It might not be that sock color causes CGPAs to change, but that sock color is associated with another cause, or causes, that do. White sock color sometimes, we might say to ourselves, is associated with athletic socks, and athletic socks are tighter fitting, and it’s this tight fit that causes (another cause) itchiness, and the itchiness sometimes causes distraction during exams. This is a loose causal chain, but an intact one.

As above, we can tell an infinite number of these cause-of-a-cause stories, the difference being that here it is much harder to keep track of the essences of the problem. Cause isn’t always so easy! Just ask physicists trying to measure effects of teeny weeny particles.

If we do not have, or can not form, a clear causal chain in our mind, we excuse ourselves by saying sock color is “correlated” or (again) “linked to” CGPA, with the understanding that cause is mixed in somehow, but we do not quite know how to say so, or at least not in every case. We know sock color is relevant (to the probability), but the only way we would keep it in the model, as said above, is if it is important to a decision we make.

Part of any decision, though, is knowledge of cause. If we knew the essences of socks, and the essence of all things associated with sock color, and we judge that these have no causal power to change CGPA, then it would not matter if there were any difference in calculated probabilities between (7a) and (7b). We would expunge sock color from our model. We’d reason that even a handful of beans tossed onto the floor can take the appearance of a President’s profile, but we’d know the pattern was in our minds and not caused intentionally by the bean-floor combination.

If we knew that, sometimes and in some but not necessarily all instances, that sock color is in the causal chain of CGPA (as in for instance tightness and itchiness) then we might include sock color in our model but only if it were important for decision.

If we ignorant (but perhaps only suspicious) of the causal chain of sock color, which for some observations in some models we will be, we keep the observation only if the decision would change.

Note carefully that it is only knowledge of cause or decision that lead to use accepting or rejecting any observable from our model. It has nothing to do (per se) with any function of measurements. Cause and decision are king in the predictive approach. Not blind algorithms.

In retrospect, this was always obvious. Even classical statisticians (and the researchers using these methods) do not put sock color into their models of grade point. Every model begins with excluding an infinity of non-causes, i.e. of observations that can be made but that are known to be causally irrelevant (if not probabilistically) irrelevant to the proposition of interest. Nobody questions this, nor should they. Yet to be perfectly consistent with classical theory, we’d have to try and “reject” the “null” hypotheses of everything under, over, around, and beyond the sun, before we were sure we found the “true” model.

Lastly, as said before and just as obvious, if we knew the cause of Y, we don’t need probability models.

Next week: real practical examples!

Homework I do not expect to “convert” those trained in classical methods. These fine folks are too used to the language in those methods to switch easily to this one. All I can ask is that people read Uncertainty for a fuller discussion of these topics. The real homework is to find an example of or try to define “linked to” without resorting somewhere to causal language.

Once you finish that impossible task, find a paper that says its results (at least in part) were “due to” chance. Now “due to” is also causal language. Given that chance is only a measure of ignorance, and therefore cannot cause anything, and using the beans-on-floor example above, explain what it is people are doing saying results were “due to” chance.

January 8, 2018 | 15 Comments

Jaws and the Meaning of Life

Stream: What the Atheist Claim of the Meaninglessness of Life Would Mean (If it were True)

Links fixed!

There is a scene early on in the killer-shark movie Jaws which has marine biologist Matt Hooper (Richard Dreyfuss) explains to Amity’s Mayor Larry Vaughn (Murray Hamilton) the nature of sharks.

“Mr Vaughn,” says Hooper, “What we are dealing with here is a perfect engine—an eating machine. It’s really a miracle of evolution. All this machine does is swim, and eat, and make little sharks. And that’s all.”

Is this explanation true? If so, then why doesn’t it also apply to ocelots? What else besides running, eating, and making little ocelots does this carnivorous beastie do?

And if it works for sharks and ocelots, why not also for dandelions, cockroaches, and ratbirds (pigeons)? And if for them, why not for all life? Why not for you, dear reader?

After all, what else do people do except scurry about, eat, and make more people?

A bag of bones

If life can be reduced to biology, to nothing but chemical and physical interactions, then the explanation that all life, including our own, is meaningless futile repetition must be true.

Don’t pass too quickly by “meaningless.” This is the main point. If our lives are solely biology, then our lives have no meaning. This is a stronger conclusion than you might think. For it follows that any meaning anybody ascribes to any event in life is itself meaningless. Any and all moral judgments are mere prejudice, the result of particular arrangements of chemicals operating under unbreakable physical laws.

If all moral judgments are prejudice, then everything anybody ever thinks or says is opinion. And it’s forced opinion, at that. All opinions are the result of chemicals pushing this way and that, forming unwilled patterns in brains, under the control of nobody.

Who asked for your opinion?

You say rape is wrong? That’s just your opinion. Worse, it’s an opinion you have no choice but to believe, since the opinion is formed in a brain operating under fixed laws. You think murder is immoral? Well, there is no such thing as immorality, and cannot be, since for acts to be moral or immoral, acts cannot be meaningless. Meaning defines morality.

So what?, you might think. Individual people might be nothing more than their biology, because what really matters is the human race itself.

But this must be false, because []

Science is not the answer

Science cannot rescue us from this bleak conclusion.

Celebrity scientist Neil deGrasse Tyson is fond of pointing out the meaningless of life.[]

The problem of evil

What we have been discussing is the Problem of Evil. If atheism is true, if life really is nothing more than biology, then []

Put some meaning in your life and click here to read the rest.

Addendum

I was having a debate with an atheist over this Stream article. He insisted that “not causing unnecessary harm” (à la Sam Harris) was the atheistic objective moral judgment sought. It isn’t, as the word “unnecessary” proves. Causing any harm, necessary or unnecessary, is without meaning, if atheism is true.