TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Automated Inference on Criminality Using Face Images

48 pointsby igonvalueover 8 years ago

11 comments

nlover 8 years ago
I thought this was a joke when I read the abstract, but it appears to be a genuine paper.<p>This paragraph in particular is one of the worst examples I&#x27;ve ever seen of researchers NOT UNDERSTANDING WHAT THEY ARE DOING:<p><i>Unlike a human examiner&#x2F;judge, a computer vision algorithm or classifier has absolutely no subjective baggages, having no emotions, no biases whatsoever due to past experience, race, religion, political doctrine, gender, age, etc., no mental fatigue, no preconditioning of a bad sleep or meal. The automated inference on criminality eliminates the variable of meta-accuracy (the competence of the human judge&#x2F;examiner) all together.</i><p>Please, read <i>Weapons of Math Destruction</i> and understand how excellent machine learning is at discovering and exploiting the biases in datasets.<p>Edit, no, sorry, it gets worse:<p><i>the upper lip curvature is on average 23.4% larger for criminals than for non-criminals.</i><p><i>the distance d between two eye inner corners for criminals is slightly shorter (5.6%) than for non-criminals</i>
评论 #12993470 未加载
评论 #12984097 未加载
评论 #12986787 未加载
评论 #12984080 未加载
评论 #12984141 未加载
评论 #12984056 未加载
评论 #12988164 未加载
评论 #12984405 未加载
评论 #12984076 未加载
sp332over 8 years ago
This would be more useful if it were applied exactly the opposite way. What facial features is the court system or even society at large biased against?
a_bonoboover 8 years ago
It looks like they didn&#x27;t split up the two training sets (criminal&#x2F;noncriminal) into two testing and training sets?<p>Which would explain this &#x27;paradox&#x27;, it&#x27;s just overtraining:<p>&gt;The seeming paradox that Sc [the criminal set] and Sn [the noncriminal set] can be classified but the average faces of Sc [the criminal set] and Sn [the noncriminal set] appear almost the same can be explained, if the data distributions of Sc [the criminal set] and Sn [the noncriminal set] are heavily mingled and yet separable.<p>They&#x27;re heavily mingled because they&#x27;re identical and you&#x27;re just testing your predictions with your training data.
评论 #12984478 未加载
ongoodieover 8 years ago
Overall I do not think that the result is surprising. The large genetic deviations result in deviant behaviour and in deviant face. On the other hand it is next to useless for law enforcement since if it is applied to general population the majority of criminal-like faces belongs to law-abiding people.<p>The fact that we do not like the result does not make it false. For validation, see page 4, where they checked that a random labeling of images does not produce such a good distinguisher.
iamthepiemanover 8 years ago
I can&#x27;t think of a comment a that doesn&#x27;t immediately invoke Godwin. This is like the setup to a Philip k Dick story. I wonder how many people outside of HN would think this is a perfectly normal result of a perfectly scientific study.
mjburnsover 8 years ago
People forget that arXiv is just an academic wikipedia. Anyone can post an &quot;article&quot; here. So a being &quot;published&quot; here is meaningless as to potential validity.<p>When referees at real journals actually do their jobs correctly, they check arXiv when given manuscripts to read &amp; reject them if they have been posted to arXiv as violating the &quot;no prior publication&quot; rules at the real journals.
Friction87over 8 years ago
From the abstract I gathered that average looking people are generally considered law-abiding whereas people with outlier features are more likely to fall into the criminal category: &quot;The variation among criminal faces is significantly greater than that of the non-criminal faces.&quot;<p>Likening this to the &quot;wage gap&quot; where the XY chromosome is responsible for more outlier behavior: both the top of society and the bottom is both heavily dominated by male participants, whereas the female population is closer to the average and has far fewer outliers. Could this be related?<p>There&#x27;s variance in XY chromosomes that cause men to swing wildly on the scale in both positive and negative directions. There seems to be an answer to the hypothesis that asserts that individuals with wildly differing attributes seem more often than not to fall on the outside of the law.
synapticaxonover 8 years ago
I can&#x27;t believe nobody has mentioned William Herbert Sheldon and his famous Somatotyping. This fell out of favor but he was systematically cataloging body features as a function of criminality.
mattnewtonover 8 years ago
So people with more average characteristics are less likely to be convicted of a crime? Could mean they are at a disadvantage with juries.
HS1over 8 years ago
Fyi<p>Evidence from Meta-Analyses of the Facial Width-to-Height Ratio as an Evolved Cue of Threat<p><a href="http:&#x2F;&#x2F;journals.plos.org&#x2F;plosone&#x2F;article?id=10.1371&#x2F;journal.pone.0132726" rel="nofollow">http:&#x2F;&#x2F;journals.plos.org&#x2F;plosone&#x2F;article?id=10.1371&#x2F;journal....</a>
aseippover 8 years ago
I haven&#x27;t read the paper, and I don&#x27;t really know much about ML, but this part stuck out to me from the abstract:<p>&gt; All four classifiers perform consistently well and produce evidence for the validity of automated face-induced inference on criminality, despite the historical controversy surrounding the topic.<p>I realize the authors are intentionally skirting around this bit (it&#x27;s not really the point of their paper), but the &quot;problem&quot; isn&#x27;t that some physical features may indicate criminality, with some level of success. That&#x27;s cool or whatever I guess, but hardly an issue or truly revolutionary I think from a social perspective (in person, people tend to understand &quot;vibes&quot; rather well, and bad vibes come from a number of things like body language or visual cues about a person. Humans have their own inference systems for these things, flawed as they are.)<p>No, the problem -- the &quot;controversy&quot; surrounding the topic -- IMO, is that, almost with 100% certainty, any implementation of this system will be completely left unchecked, will effectively be private, and will be totally unaccountable by any practical means.<p>Do the authors of this paper really think any implementation of this system would be open to the public in any accountable way, if used by say, LEOs? You know, as opposed to it being a big &quot;every-criminal.sql&quot; dump, based on hoarded data mining, and driven along and utilized by proprietary algorithms, created by some company selling to governments? LEOs in places like the US have already shown their hands with strategies like parallel reconstruction and the downright willingness to fabricate evidence out of thin air.<p>Really, who cares what some data science nerds think of their fancy criminal face models, and whether they think they&#x27;re &quot;accurate despite the controversy&quot;, when the police can just say &quot;It&#x27;s accurate, I say so, you&#x27;re going to jail&quot; and they can completely make shit up to support it? It&#x27;s not a matter of whether the actual thing is accurate, it&#x27;s whether or not it gives them a reason to do whatever they like.<p>It reminds me of rhetoric people said, about building walls around Mexico wrt the election. That can&#x27;t happen. Who would build it. It&#x27;d be huge. Hard. Realistically? It&#x27;d be &quot;easy&quot;. Humans have been building walls for a long, long time. It&#x27;s not unthinkable. The difficult part is actually murdering people who would try to cross the wall by gunning them down -- and they will try to cross. I mean, you probably don&#x27;t have to kill <i>too many</i> people to send the message. Just, enough of them. The Iron Curtain was a real thing, too, after all.<p>This is similar. The algorithm is the &quot;easy&quot; part. It&#x27;s &quot;only&quot; some science. No, the hard part is dealing with the consequences. The hard part is closing the box of Pandora after you opened it.