ClaimsYann LeCun

“A big advantage of ConvNets for detection is their computational efficiency:”

Yann LeCun, co-author of paper 2010 Convolutional Networks and Applications in Vision (yann.lecun.com) archived copy

Factual AI capabilities

A big advantage of ConvNets for detection is their computational efficiency.

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.

What would count as true or false

Judge against knowledge as of 2010. TRUE: the technical literature up to 2010 shows that ConvNets used for detection are computationally efficient, notably because convolutions applied over a whole image share computation across overlapping windows, and this is widely recognised as a significant advantage over competing detectors. MOSTLY TRUE: efficiency is a real advantage, but it holds mainly relative to naive sliding-window classifiers rather than to all competing detection methods of the time. MIXED: ConvNets were efficient in some detection settings but clearly slower than competing methods in others. MOSTLY FALSE: the efficiency gain was marginal or applied only in narrow cases. FALSE: ConvNet detectors of the time were generally less efficient computationally than the alternatives.

Set before the research began.

Evidence (3)

  1. E1 Robust Real-Time Face Detection International Journal of Computer Vision (Viola and Jones; hosted at RWTH Aachen University) · 2004

    “The approach was used to construct a face detection system which is approximately 15 times faster than any previous approach.”
  2. E2 Pedestrian Detection withUnsupervised Multi-Stage Feature Learning ar5iv (arXiv 1212.0142) Own record

    “While processing speed is not the focus of this paper, features and classifier approximations introduced by dollar-bmvc-10 and benenson2012pedestrian may be applicable to deep learning models for faster detection, in addition to GPU optimizations.”
  3. E3 Object Detection with Sliding Windows | Shav Wiki Shav Wiki · 1 January 2014

    “The convolutional implementation of sliding windows allows these windows in the convnet to share a lot of computation.”

Reasoning

A later secondary source says convolutional sliding windows share computation [E3], and a 2004 face detector was reported as approximately 15 times faster than earlier approaches [E1]. However, no evidence compares ConvNet detectors' efficiency with that of other detectors as of 2010, so the evidence cannot separate the true, mostly true and mixed outcomes under the criteria.

Cite this claim

Yann LeCun, “A big advantage of ConvNets for detection is their computational efficiency.” (said 2010). Verdict: unresolvable. Reliability API, claim c_bd82e9124139, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_bd82e9124139

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.