Scores by kind of claim
Each kind of claim is scored on its own, out of 100: the share that held up, with the range it likely falls in. Fewer resolved claims give a wider range.
-
Factual claims
Statements about the present or the past, checked against the record: did they hold up?
78 72 83 -
Predictions
Statements about events outside the person's control, checked once they fell due: did they come true?
64 43 83 -
Promises
Commitments to the person's or organisation's own future action, with something to deliver: was it delivered?
Insufficient record: 0 promises resolved, 5 needed for a score.
All types combined: 77 (likely 71 to 82, high confidence), from 84 claims: 79 factual claims, 5 predictions. It mixes kinds of claim in different proportions for different people, so compare people kind by kind.
Yann LeCun's claims
55 claims
“Using this loss with GW (X) = W T X and a regularizer of the form ||W ||2 gives the familiar linear support vector machine.”
LeCun et al., tutorial authors 2006 A Tutorial on Energy-Based Learning (yann.lecun.com)
The hinge loss with G_W(X)=W^T X and a ||W||^2 regularizer gives the linear support vector machine.
Mostly true True in substance, wrong in a detail.
The second judge would not resolve this claim; the verdict shown is the first judge's.
The evidence is the person's or their organisation's own records.
What would count as true or false
Judge as a mathematical fact. True: minimising the hinge loss on a linear score W^T X (using the margin or label formulation in LeCun et al.'s 2006 energy-based learning tutorial) plus an L2 penalty ||W||^2 gives the standard soft-margin linear SVM objective. Mostly true: the equivalence holds up to minor details (a bias term, scaling constants, or a margin parameter). Mixed: it holds only under substantial extra assumptions. Mostly false: the resulting objective differs materially from the SVM. False: the formulation does not correspond to a linear SVM.
Evidence
- Yann LeCun NYU Courant Institute (talk slides, author's own) Own record
“Linear Model + Margin Loss + Regularization = SVM”
- Yann LeCun NYU Courant Institute (talk slides, author's own) Own record
“Minimize the Regularization term: Make W as short as possible.”
- Yann LeCun NYU Courant Institute (talk slides, author's own) Own record
“You get a Support Vector Machine.”
Reasoning
The criteria ask for a judgement on the mathematics. Minimising the hinge loss max(0, m − Y·WᵀX) summed over samples, plus λ||W||², is the soft-margin linear SVM objective up to how the constants are scaled. The match is not exact because G_W(X)=WᵀX has no separate bias term, which the standard SVM includes. Under the criteria that is a minor detail, so the verdict is mostly true rather than true. LeCun's own slides (E1, E2, E3) say the same thing ("Linear Model + Margin Loss + Regularization = SVM"), but they are not independent, so the verdict rests on checking the mathematics, not on the slides.
Cite this claim
Yann LeCun, “The hinge loss with G_W(X)=W^T X and a ||W||^2 regularizer gives the linear support vector machine.” (said 2006). Verdict: mostly true. Reliability API, claim c_28b6f7d1d2e2, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_28b6f7d1d2e2
Claims without a verdict
- Unresolvable 110
- Not yet due 3
- Not researched 309
- Not checkable 669
Is a verdict here wrong, or do you speak for Yann LeCun? Reply on the claim's own page, with the form at its foot (how replies work). Verdicts are made automatically; how verdicts and scores are made.