Skip to content

fix: correct modified Huber loss gradient in ml/incr/binary-classification #15317

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Abhist17 wants to merge 3 commits into
base: develop
Choose a base branch
from
Open

fix: correct modified Huber loss gradient in ml/incr/binary-classification #15317

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
63784cb
fix: correct modified Huber loss gradient in `ml/incr/binary-classifi…
Abhist17 Sep 18, 2026
1f23797
test: consolidate the modified Huber loss tests
Abhist17 Sep 29, 2026
61f5267
test: simplify case comments
Abhist17 Oct 1, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions lib/node_modules/@stdlib/ml/incr/binary-classification/lib/model.js
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -341,8 +341,8 @@
d = y * this._dot( x.data, x.strides[ 0 ], x.offset );
if ( d < -1.0 ) {
this._add( x, 4.0*eta*y );
} else {
this._add( x, eta*( y-(d*y) ) );
} else if ( d < 1.0 ) {
this._add( x, 2.0*eta*( y-(d*y) ) );
}
Comment on lines 342 to 346

@nakul-krishnakumar nakul-krishnakumar Sep 28, 2026 •
edited
Loading

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This does not look correct to me as here else if comes after an else. According to sklearn, the correct branching should be:

if ( d >= 1.0 ) {
    // no update
} else if ( d >= -1.0 ) {
    this._add( x, 2.0*eta*( y-(d*y) ) );
} else {
    this._add( x, 4.0*eta*y );
}

Reference

To make it simpler we can do this:

if ( d < -1.0 ) {
    this._add( x, 4.0*eta*y );
} else if ( d < 1.0 ) {
    this._add( x, 2.0*eta*( y-(d*y) ) );
}

@nakul-krishnakumar nakul-krishnakumar Sep 28, 2026

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Another reference is our @stdlib/ml/base/loss/float64/modified-huber-gradient implementation.

return this;
});
Expand All @@ -357,7 +357,7 @@
* [@shalevshwartz:2011a]: https://doi.org/10.1007/s10107-010-0420-4
*
* @private
* @name _pegasos

Check warning on line 360 in lib/node_modules/@stdlib/ml/incr/binary-classification/lib/model.js

View workflow job for this annotation

GitHub Actions / Lint Changed Files 

Unknown word: "pegasos"
* @memberof Model.prototype
* @type {Function}
* @returns {number} learning rate
Expand Down
41 changes: 40 additions & 1 deletion lib/node_modules/@stdlib/ml/incr/binary-classification/test/test.model.js
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,8 @@
// MODULES //

var tape = require( 'tape' );
var array = require( '@stdlib/ndarray/array' );
var Float64Array = require( '@stdlib/array/float64' );
var Model = require( './../lib/model.js' );


Expand All @@ -32,4 +34,41 @@ tape( 'main export is a function', function test( t ) {
t.end();
});

// TODO: add tests
tape( 'the function supports modified Huber loss', function test( t ) {
var expected;
var model;
var opts;
var eta;
var x;

eta = 0.1;
opts = {
'lambda': 0.0,
'learningRate': [ 'constant', eta ],
'loss': 'modifiedHuber',
'intercept': false
};
x = array( new Float64Array( [ 1.0 ] ) );

// Case 1: `-1 <= y*f(x) < 1` (quadratically smoothed region):
model = new Model( 1, opts );
model._weights[ 0 ] = 0.3; // eslint-disable-line no-underscore-dangle
model.update( x, 1 );
expected = 0.3 + ( 2.0*eta*( 1.0-(0.3*1.0) ) );
t.strictEqual( model._weights[ 0 ], expected, 'returns expected value' ); // eslint-disable-line no-underscore-dangle

// Case 2: `y*f(x) >= 1`, where the gradient is zero and the weights are unchanged:
model = new Model( 1, opts );
model._weights[ 0 ] = 2.0; // eslint-disable-line no-underscore-dangle
model.update( x, 1 );
t.strictEqual( model._weights[ 0 ], 2.0, 'returns expected value' ); // eslint-disable-line no-underscore-dangle

// Case 3: `y*f(x) < -1`:
model = new Model( 1, opts );
model._weights[ 0 ] = -2.0; // eslint-disable-line no-underscore-dangle
model.update( x, 1 );
expected = -2.0 + ( 4.0*eta*1.0 );
t.strictEqual( model._weights[ 0 ], expected, 'returns expected value' ); // eslint-disable-line no-underscore-dangle

t.end();
});
Loading