The first commercially viable check reading system, deployed by AT&T/NCR in 1996, used a ConvNet rather than an SVM.
Mostly true True in substance, wrong in a detail.
Factual: Statements about the present or the past, checked against the record: did they hold up?
What would count as true or false
Judged as of 15 May 2014. True: historical records (LeCun et al. 1998 'Gradient-based learning applied to document recognition' and accounts by AT&T/NCR) show that a check-reading system deployed commercially by AT&T/NCR around 1996 used a convolutional network (LeNet) for character recognition, not an SVM, and that it is reasonably described as the first commercially viable such system. Mostly true: the system used a ConvNet and was deployed by AT&T/NCR, but the deployment year differs slightly (within about a year), or the claim that it was 'first' is debatable. Mixed: it used a ConvNet, but neither the 'first commercially viable' description nor the date and deployer hold. Mostly false: the system used a ConvNet only as a minor part, or the claim that it was first is clearly contradicted by earlier commercial systems. False: the system did not use a ConvNet, or used an SVM.
Set before the research began.
Evidence (4)
-
E1 Convolutional neural network - Wikipedia Wikipedia · 31 August 2013
“The system was integrated in NCR's check reading systems, and fielded in several American banks since June 1996, reading millions of checks per day.”
-
E2 Convolutional neural network - Wikipedia Wikipedia · 31 August 2013
“LeNet-5, a pioneering 7-level convolutional network by LeCun et al. in 1995,[56] classifies hand-written numbers on checks digitized in 32×32 pixel images.”
-
E3 Global training of document processing systems using graph transformer networks SciSpace (hosted copy of Bottou, Bengio & LeCun, AT&T Labs) Own record
“The recognizer used in the Check Reading System is a convolutional neural network coined LeNet5.”
-
E4 ConvNets Patent by AT&T Bell Labs Flying Penguin (blog) · 7 July 2021 Own record
“Commercial deployment in banks started in 1995.”
Reasoning
Wikipedia, an independent source, says the convolutional network system was built into NCR's check reading systems and used in American banks from June 1996 (E1). It also identifies LeNet-5, a convolutional network, as the network that read checks (E2). The AT&T Labs paper co-authored by LeCun likewise names a convolutional network, LeNet5, as the recognizer in the check reading system and does not mention an SVM (E3). So the ConvNet, the deployer and the date are supported. However, no source here shows it was the "first commercially viable" system, and E4 says commercial deployment in banks started in 1995, so the date and the "first" wording are debatable.
Cite this claim
Yann LeCun, “The first commercially viable check reading system, deployed by AT&T/NCR in 1996, used a ConvNet rather than an SVM.” (said 15 May 2014). Verdict: mostly true. Reliability API, claim c_60030e574946, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_60030e574946
This claim is in dataset release v1.1.2, with the same verdict.
Reply to this claim
More factual claims by Yann LeCun
-
Mostly true
“For starters, transformers are equivariant to permutations.”
Transformers are equivariant to permutations, as ConvNets are equivariant to translations.
-
Mostly true
“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.”
-
Mostly true
“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.”
-
Mostly true
“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.”
-
Mostly true
“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…”
Verdicts are made automatically; how verdicts are made.