[nakul-krishnakumar]

---

type: pre_commit_static_analysis_report
description: Results of running static analysis checks when committing changes. report:

• task: lint_filenames status: passed
• task: lint_editorconfig status: passed
• task: lint_markdown status: passed
• task: lint_package_json status: passed
• task: lint_repl_help status: passed
• task: lint_javascript_src status: passed
• task: lint_javascript_cli status: na
• task: lint_javascript_examples status: passed
• task: lint_javascript_tests status: passed
• task: lint_javascript_benchmarks status: passed
• task: lint_python status: na
• task: lint_r status: na
• task: lint_c_src status: passed
• task: lint_c_examples status: passed
• task: lint_c_benchmarks status: passed
• task: lint_c_tests_fixtures status: na
• task: lint_shell status: na
• task: lint_typescript_declarations status: passed
• task: lint_typescript_tests status: passed
• task: lint_license_headers status: passed ---

Resolves None.

Description

What is the purpose of this pull request?

This pull request:

• Adds ml/base/loss/float64/hinge.

Related Issues

Does this pull request have any related issues?

This pull request has the following related issues:

• None.

Questions

Any questions for reviewers of this pull request?

No.

Other

Any other information relevant to this pull request? This may include screenshots, references, and/or implementation notes.

No.

Checklist

Please ensure the following tasks are completed before submitting this pull request.

• Read, understood, and followed the contributing guidelines.

AI Assistance

When authoring the changes proposed in this PR, did you use any kind of AI assistance?

• Yes
• No

If you answered "yes" above, how did you use AI assistance?

• Code generation (e.g., when writing an implementation or fixing a bug)
• Test/benchmark generation
• Documentation (including examples)
• Research and understanding

Disclosure

If you answered "yes" to using AI assistance, please provide a short disclosure indicating how you used AI assistance. This helps reviewers determine how much scrutiny to apply when reviewing your contribution. Example disclosures: "This PR was written primarily by Claude Code." or "I consulted ChatGPT to understand the codebase, but the proposed changes were fully authored manually by myself.".

None.

---

@stdlib-js/reviewers

[stdlib-bot]

Coverage Report

Package	Statements	Branches	Functions	Lines
ml/base/loss/float64/hinge	$\\color{green}194/194$ $\\color{green}+0.00\\%$	$\\color{green}11/11$ $\\color{green}+0.00\\%$	$\\color{green}2/2$ $\\color{green}+0.00\\%$	$\\color{green}194/194$ $\\color{green}+0.00\\%$

The above coverage report was generated for the changes in this PR.

[nakul-krishnakumar]

Clarifications required:

1. hinge( y, p ) loss function expect y to be either +1 or -1. Should I mention this in the README.md and repl.txt?
2. Please have a look at the fixture generation and lmk if the approach is apt.
3. Should I keep y as integer or double? y is the target class label, so it would be an integer, but keeping it as double seems more appropriate.

[nakul-krishnakumar]

/stdlib update-copyright-years

[kgryte]

Left initial comments.

[nakul-krishnakumar]

Before this gets merged, I wanted to discuss a possibility. If we were supposed to add a threshold arg which would be either 1.0 or 0.0, then we could reuse this implementation for perceptron-loss. Reference: sklearn

[kgryte]

@nakul-krishnakumar Yes, fine with adding a threshold argument. Although, it need not be restricted to 0 and 1. Can be any value greater than or equal to 0.

[nakul-krishnakumar]

/stdlib merge

[kgryte]

Missing parameter list.

[kgryte]

Why "between two double-precision floating-point numbers"? Why not "with respect to a model parameter" as you did for hinge-gradient? Question applies throughout this PR.

[nakul-krishnakumar]

Here, “with respect to a model parameter” would not be applicable, since we are not differentiating the loss function with respect to a model parameter, as we would when computing the loss gradient. However, I’m open to changing the description to something more appropriate if you have a better suggestion.

[nakul-krishnakumar]

To make things more clear:

1. Hinge Loss function

$$L = \max(0, 1-yf(x))$$

2. Hinge loss function gradient

$$\frac{\partial L}{\partial w} = \begin{cases} -yx, & y(w^Tx+b)<1\\[4pt] 0, & y(w^Tx+b)>1 \end{cases}$$

Here it is being differentiated wrt to model parameter.

[kgryte]

@nakul-krishnakumar Would you be up for moving this PR along? Cheers!

[0PrashantYadav0]

Suggested change

[0PrashantYadav0]

Suggested change

	v = hinge( 2.0, 5.0, NaN );
	v = hinge( 2.0, 1.0, NaN );

y = 5.0 already returns NaN, so this example doesn't show the NaN case for p.

[0PrashantYadav0]

Suggested change

	<!-- /.usage -->
	<!-- /.usage -->
	<section class="notes">
	## Notes
	- When `t = 1.0`, we get the loss used by SVM; whereas, when `t = 0.0`, we get the loss used by the Perceptron.
	</section>
	<!-- /.notes -->

Same note as hinge-gradient. t was added so this package could also serve as the Perceptron loss.

[0PrashantYadav0]

Suggested change

	- **t**: `[in] double` threshold value.
	- **t**: `[in] double` margin threshold.

Matches line 63 and hinge-gradient. Same in lib/main.js:32, lib/native.js:32, docs/types/index.d.ts:28, src/main.c:27 and docs/repl.txt:12 (Margin threshold.).

[0PrashantYadav0]

Suggested change

	*
	*
	* ## Notes
	*
	* - If `y` is not +1 or -1, the function returns `NaN`.
	* - When `t = 1.0`, we get the loss used by SVM, whereas when `t = 0.0`, we get the loss used by the Perceptron.
	*

Matches hinge-gradient/lib/main.js. Add the SVM/Perceptron note in docs/types/index.d.ts:26 and after docs/repl.txt:7 as well.

[0PrashantYadav0]

Suggested change

	* var v = hinge( 5.0, -1.0, 0.234 );
	* var v = hinge( 5.0, -1.0, 0.234 );

Same in lib/native.js:50, lib/index.js:38 and docs/types/index.d.ts:46.