Just came across this :

https://www.telegraph.co.uk/news/2022/1 ... gniter-rhr

Check it out , interesting for at least one member!

Sorry, don't take it personally.

“You can’t cheat the algorithm”
Well that’s bullshit for a start.
The algorithms are known to be flawed as they are affected by the biases of developers. There’s plenty of research on the flaws.

https://www.unimelb.edu.au/newsroom/new ... telligence

https://www.media.mit.edu/articles/faci ... her-finds/

https://createdigital.org.au/ai-reflect ... echnology/

"Well that’s bullshit for a start."

Someone can breath easy then.

Fucking paywall shit.

mannanman wrote: ↑Mon Nov 07, 2022 9:19 pm Fucking paywall shit.

The link in OP? I was able to read it.

Lord Lucan 'is exact match for Australian pensioner', facial recognition expert claims
Algorithm that found killers of Jamal Khashoggi identifies Brisbane man, 87, as missing British aristocrat

Lord Lucan and a pensioner in Australia have been matched as “the same individual” by a facial recognition expert who unmasked the Skripal poisoners and Jamal Khashoggi’s killers.

Professor Hassan Ugail believes his world-leading AI photo analysis technology may have cracked Britain’s most infamous murder mystery as a matter of “science and mathematical fact”.

Lord Lucan, an Eton-educated aristocrat, went on the run in 1974 after 29-year-old Sandra Rivett, who worked for him, was bludgeoned to death with a lead pipe in his Belgravia home.

His disappearance without a trace captured the public imagination like few others and led to decades of speculation about his fate.

Detectives investigating the mystery found three Cluedo cards in the missing Earl’s car, prompting suggestions the infamous murder of his nanny may have been pre-planned, it was revealed on Friday.

Ms Rivett’s son, Neil Berriman, has spent years trying to establish the whereabouts of the man that an inquest jury concluded in 1975 was responsible for his mother’s death

After becoming convinced an 87-year-old man who lives near Brisbane, Australia, is actually Lord Lucan, Mr Berriman asked Professor Ugail to compare three pictures of the pensioner to four old pictures of the vanished peer.

His facial recognition technology, developed over 15 years and trained using millions of facial images, provides “very, very detailed” analysis of a picture that “goes far deeper than what the human eye can see”.

The results were nothing short of extraordinary.

'You can't cheat the algorithm'

“According to the computer algorithm, based on thousands of experiments, these pictures belong to the same individual or someone who looks extremely like them -  like identical twins,” Professor Ugail said.

“This is science and mathematical fact. You can’t cheat the algorithm.”

The computer scientist, based at the University of Bradford, has made global headlines for his work identifying some of the world’s most wanted men.

In 2018, he helped the investigative website Bellingcat unmask the two Russian agents responsible for poisoning the former spy Sergei Skripal and his daughter in the town of Salisbury.

He has also identified three suspects linked to the killing of the Saudi journalist Jamal Khashoggi in 2018 and helped to uncover an allaged Nazi war criminal last year.

Professor Ugail claimed his computer algorithm has “never been wrong”.

It will analyse two photos and provide a “similarity index”, or a percentage likelihood of a match.

Anything above 75 per cent is considered to be either the same individual or someone who looks identical to them, such as a twin, because the different way images are taken and their pixel compilation means a 100 per cent match is impossible.

For all the pictures comparing Lord Lucan to the Australian pensioner, the similarity index was above 75 per cent and in most cases over 80 per cent.

'Thousands of dimensions'

Professor Ugail said: “As humans, we see objects like faces in three dimensions but a computer can see them in thousands of dimensions. 

“By using millions of images of faces as data, we have trained a very powerful computer algorithm to recognise and identify a face – even within a database of millions of faces and even if the input face is only a partial blurry image of a face.

Unlike humans, the computer algorithm looks at a face very deeply - from the colour, texture, the different shapes, the ratios and proportions. It’s a very, very detailed way of analysing an image which goes far deeper than what the human eye can see.”

Of the results of his Lord Lucan analysis, he added: “In my experience, over the thousands of experiments we have done over many years, involving millions and millions of faces, the face recognition algorithm would not report such close resemblance unless it is from the same individual, or someone who looks extremely like them, such as their identical twin.”

A separate company specialising in facial recognition ran the same tests and came to the same conclusion, the Mirror reported.

Mr Berriman told the newspaper: “I’ve spent nine years trying to prove this man is Lucan. Now, with this new scientific information, the police must act.

“This isn’t emotion. It’s fact.”