Description
Return a uniformly smoothed binary target pair.
Training losses compare predicted probabilities with target mass, and temperature rescales logits before a probability transformation. Label smoothing and focal weighting alter particular loss terms; evaluating one term or transformation does not train a model, select its hyperparameters or establish calibrated confidence.
Return a uniformly smoothed binary target pair.
When to use this tool
Use Classification: Binary Label Smoothing Target Pair when evaluating the stated loss contribution, smoothed target or temperature transformation.
How to use the tool and how it works
- Enter label, smoothing. Use the units and conventions below.
- Run the tool. Correct any format, range, or cross-field input error before interpreting the output.
- Read result. Expand result to inspect the returned fields.
Return a uniformly smoothed binary target pair. Require label zero/one and smoothing in [0,1]; each target combines its one-hot mass times 1-smoothing with smoothing/2.1
Uniform label smoothing mixes a one-hot target with equal mass over every class, including the true class.2
| Input field | Meaning and units | Accepted input |
|---|---|---|
| label | Label | Required. Safe integer. Calculation range: 0 through 1, inclusive. |
| smoothing | Smoothing | Required. Finite number. |
| Output field | Interpretation | Returned shape |
|---|---|---|
| result | Returned calculated value or nested object under the stated output convention. | Required. Object. |
| result.negativeTarget | Negative Target | Calculated result member: number. |
| result.positiveTarget | Positive Target | Calculated result member: number. |
{
"input": {
"label": 1,
"smoothing": 0.2
},
"output": {
"result": {
"negativeTarget": 0.1,
"positiveTarget": 0.9
}
}
}Limitations and assumptions
- At smoothing one both targets are 0.5; no loss, training or recommended smoothing level is determined.
- Outputs are nested inside result. Inputs must satisfy both the field types and the cross-field conditions above. Invalid input causes an error; a successful false flag or documented placeholder is a different result.
- Numeric calculations use binary floating point unless exact decimal-integer arithmetic is explicitly stated. No hidden measurement tolerance or decimal rounding is added; nonfinite or out-of-contract results are rejected. Numeric signed-zero results are exported as zero.
Alternative or Complementary methods
Sum the appropriate contributions for a complete loss, and use a validated training/calibration procedure and held-out evaluation to select hyperparameters and assess predictions.
References
-
Calculation and input contract: classification/binary-label-smoothing-target-pair — tool.io
-
Rethinking the Inception Architecture for Computer Vision — Christian Szegedy and coauthors
Similar or alternative tools
- Binary Logit Temperature Slope
Return temperature-scaled logistic probability and its logit derivative.
- Multiclass Cross Entropy Smoothing Term
Return one label-smoothed cross-entropy class term.