ClaimsYann LeCun

“The idea goes back to papers from Geoff Hinton’s lab and Yann LeCun’s group in the early 1990s”

Meta AI blog post, byline VP and Chief AI Scientist (LeCun) 4 March 2021 Self-supervised learning: The dark matter of intelligence (Meta AI)

Factual AI capabilities

The Siamese network / joint embedding idea goes back to papers from Geoff Hinton's lab and Yann LeCun's group in the early 1990s.

Unresolvable The public record could not settle it.

Factual: Statements about the present or the past, checked against the record: did they hold up?

The second judge would not resolve this claim; the verdict shown is the first judge's.

Said for an organisation they lead or work for; it counts on both records.

What would count as true or false

Judge as of 2021-03-04. True: early-1990s papers from Hinton's lab (e.g. Becker & Hinton 1992) and LeCun's group (e.g. Bromley et al. 1993/1994, Siamese networks for signature verification) introduced Siamese/joint-embedding ideas, and no clearly earlier origin is well documented. Mostly true: both lines of work exist from about 1992–1994, but there are some earlier antecedents, so 'goes back to' is slightly loose. Mixed: only one of the two groups originated the idea in the early 1990s. Mostly false: the idea clearly began substantially earlier or elsewhere, and these papers were only later contributions. False: neither group published such work in the early 1990s.

Set before the research began.

Evidence (4)

  1. E1 Self-organizing neural network that discovers surfaces in random-dot stereograms - Nature Nature · 1992

    “Becker, S., Hinton, G. Self-organizing neural network that discovers surfaces in random-dot stereograms. Nature 355, 161–163 (1992).”

  2. E2 [PDF] Signature Verification using a "Siamese" Time Delay Neural Network Advances in Neural Information Processing Systems (NeurIPS) / AT&T Bell Laboratories · 1993

    “The Siamese network has two input fields to compare two patterns and one output”

  3. E3 [PDF] Signature Verification using a "Siamese" Time Delay Neural Network Advances in Neural Information Processing Systems (NeurIPS) / AT&T Bell Laboratories · 1993

    “A similar Siamese architecture was independently proposed for fingerprint identification by Baldi and Chauvin (1992).”

  4. E4 JEPA: Yann LeCun's 30-Year Bet Against Predicting Pixels apolo.us · 15 July 2026

    “The history is usually started in 1993, but a year earlier, in 1992, Suzanna Becker and Geoffrey Hinton published a paper in Nature”

Reasoning

The evidence shows a 1992 Becker and Hinton Nature paper on a self-organizing network for random-dot stereograms [E1] and a 1993 AT&T Bell Laboratories paper describing a Siamese network [E2]. It does not name the 1993 paper's authors, so it does not link that paper to LeCun's group, and it does not describe the 1992 paper as a Siamese or joint-embedding design. The criteria's grades cannot be told apart without these links.

Cite this claim

Yann LeCun, “The Siamese network / joint embedding idea goes back to papers from Geoff Hinton's lab and Yann LeCun's group in the early 1990s.” (said 4 March 2021). Verdict: unresolvable. Reliability API, claim c_16ed62ebb064, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_16ed62ebb064

This claim is in dataset release v1.1.2, with the same verdict.

Reply to this claim

For Yann LeCun, or someone speaking for them: a correction or counter-evidence. Replies are read before anything is shown; an approved reply shows your name, role, date, statement and link. Your email address is never shown and is deleted 30 days after the decision (privacy).

More factual claims by Yann LeCun

  1. Mostly true Factual 18 March 2024
    “For starters, transformers are equivariant to permutations.”

    Transformers are equivariant to permutations, as ConvNets are equivariant to translations.

  2. Mostly true Factual 14 March 2024
    “If it were the case, we would have AI systems that could teach themselves to drive a car in 20 hours of practice, like any 17 year-old.”

    There are currently no AI systems that can teach themselves to drive a car in about 20 hours of practice, as a 17-year-old can.

  3. Mostly true Factual 8 March 2024
    “The architecture is good, the architecture of the encoder is good, but the fact that you train the system to reconstruct images does not lead it to produce to long, good generic features of images.”

    Training an encoder to reconstruct images does not lead it to produce good generic image features, though the encoder architecture is good.

  4. Mostly true Factual 8 March 2024
    “LLMs can do none of those or they can only do them in a very primitive way and they don’t really understand the physical world.”

    Current LLMs can do none of understanding the physical world, persistent memory, reasoning and planning, or only in a very primitive way.

  5. Mostly true Factual 8 March 2024
    “No, there’s one thing that autoregressive LLMs or that LLMs in general, not just the autoregressive one, but including the bird style bidirectional ones, are exploiting and its self supervised…”

    LLMs, including BERT-style bidirectional ones, exploit self-supervised learning, and LeCun has been a very strong advocate of self-supervised learning for many years.

All of them

Verdicts are made automatically; how verdicts are made.