svm_classifier.predict
fn predict(ref self: SVMClassifier<T>, X: Tensor<T>) -> (Span<usize>, Tensor<T>);Support Vector Machine classification.
Args
self: SVMClassifier - A SVMClassifier object.X: Input 2D tensor.
Returns
N Top class for each point
The class score Matrix for each class, for each point. If prob_a and prob_b are provided they are probabilities for each class, otherwise they are raw scores.
Type Constraints
SVMClassifier and X must be fixed points
Examples
fn example_svm_classifier_noprob_linear_sv_none() -> (Span<usize>, Tensor<FP16x16>) {
let coefficients: Span<FP16x16> = array![
FP16x16 { mag: 50226, sign: false },
FP16x16 { mag: 5711, sign: false },
FP16x16 { mag: 7236, sign: false },
FP16x16 { mag: 63175, sign: true }
]
.span();
let kernel_params: Span<FP16x16> = array![
FP16x16 { mag: 8025, sign: false },
FP16x16 { mag: 0, sign: false },
FP16x16 { mag: 196608, sign: false }
]
.span();
let kernel_type = KERNEL_TYPE::LINEAR;
let prob_a: Span<FP16x16> = array![].span();
let prob_b: Span<FP16x16> = array![].span();
let rho: Span<FP16x16> = array![FP16x16 { mag: 146479, sign: false }].span();
let support_vectors: Span<FP16x16> = array![
FP16x16 { mag: 314572, sign: false },
FP16x16 { mag: 222822, sign: false },
FP16x16 { mag: 124518, sign: false },
FP16x16 { mag: 327680, sign: false },
FP16x16 { mag: 196608, sign: false },
FP16x16 { mag: 104857, sign: false },
FP16x16 { mag: 294912, sign: false },
FP16x16 { mag: 150732, sign: false },
FP16x16 { mag: 85196, sign: false },
FP16x16 { mag: 334233, sign: false },
FP16x16 { mag: 163840, sign: false },
FP16x16 { mag: 196608, sign: false }
]
.span();
let classlabels: Span<usize> = array![0, 1].span();
let vectors_per_class = Option::Some(array![3, 1].span());
let post_transform = POST_TRANSFORM::NONE;
let mut classifier: SVMClassifier<FP16x16> = SVMClassifier {
classlabels,
coefficients,
kernel_params,
kernel_type,
post_transform,
prob_a,
prob_b,
rho,
support_vectors,
vectors_per_class,
};
let mut X: Tensor<FP16x16> = TensorTrait::new(
array![3, 3].span(),
array![
FP16x16 { mag: 65536, sign: true },
FP16x16 { mag: 52428, sign: true },
FP16x16 { mag: 39321, sign: true },
FP16x16 { mag: 26214, sign: true },
FP16x16 { mag: 13107, sign: true },
FP16x16 { mag: 0, sign: false },
FP16x16 { mag: 13107, sign: false },
FP16x16 { mag: 26214, sign: false },
FP16x16 { mag: 39321, sign: false },
]
.span()
);
return SVMClassifierTrait::predict(ref classifier, X);
}
// >>> ([0, 0, 0],
// [[-2.662655, 2.662655],
// [-2.21481, 2.21481],
// [-1.766964, 1.766964]])
fn example_svm_classifier_binary_softmax_fp64x64() -> (Span<usize>, Tensor<FP64x64>) {
let coefficients: Span<FP64x64> = array![
FP64x64 { mag: 18446744073709551616, sign: false },
FP64x64 { mag: 18446744073709551616, sign: false },
FP64x64 { mag: 18446744073709551616, sign: false },
FP64x64 { mag: 18446744073709551616, sign: false },
FP64x64 { mag: 18446744073709551616, sign: true },
FP64x64 { mag: 18446744073709551616, sign: true },
FP64x64 { mag: 18446744073709551616, sign: true },
FP64x64 { mag: 18446744073709551616, sign: true }
]
.span();
let kernel_params: Span<FP64x64> = array![
FP64x64 { mag: 7054933896252620800, sign: false },
FP64x64 { mag: 0, sign: false },
FP64x64 { mag: 55340232221128654848, sign: false }
]
.span();
let kernel_type = KERNEL_TYPE::RBF;
let prob_a: Span<FP64x64> = array![FP64x64 { mag: 94799998099962986496, sign: true }].span();
let prob_b: Span<FP64x64> = array![FP64x64 { mag: 1180576833385529344, sign: false }].span();
let rho: Span<FP64x64> = array![FP64x64 { mag: 3082192501545631744, sign: false }].span();
let support_vectors: Span<FP64x64> = array![
FP64x64 { mag: 3528081300248330240, sign: false },
FP64x64 { mag: 19594207602596118528, sign: true },
FP64x64 { mag: 9235613999318433792, sign: false },
FP64x64 { mag: 10869715877100519424, sign: true },
FP64x64 { mag: 5897111318564962304, sign: true },
FP64x64 { mag: 1816720038917308416, sign: false },
FP64x64 { mag: 4564890528671334400, sign: false },
FP64x64 { mag: 21278987070814027776, sign: true },
FP64x64 { mag: 7581529597213147136, sign: false },
FP64x64 { mag: 10953113834067329024, sign: true },
FP64x64 { mag: 24318984989010034688, sign: true },
FP64x64 { mag: 30296187483321270272, sign: true },
FP64x64 { mag: 10305112258191032320, sign: false },
FP64x64 { mag: 17005441559857987584, sign: true },
FP64x64 { mag: 11555205301925838848, sign: false },
FP64x64 { mag: 2962701975885447168, sign: true },
FP64x64 { mag: 11741665981322231808, sign: true },
FP64x64 { mag: 15376232508819505152, sign: false },
FP64x64 { mag: 13908474645692022784, sign: false },
FP64x64 { mag: 7323415394302033920, sign: true },
FP64x64 { mag: 3284258824352956416, sign: true },
FP64x64 { mag: 11374683084831064064, sign: true },
FP64x64 { mag: 9087138148126818304, sign: false },
FP64x64 { mag: 8247488946750095360, sign: false }
]
.span();
let classlabels: Span<usize> = array![0, 1].span();
let vectors_per_class = Option::Some(array![4, 4].span());
let post_transform = POST_TRANSFORM::SOFTMAX;
let mut classifier: SVMClassifier<FP64x64> = SVMClassifier {
classlabels,
coefficients,
kernel_params,
kernel_type,
post_transform,
prob_a,
prob_b,
rho,
support_vectors,
vectors_per_class,
};
let mut X: Tensor<FP64x64> = TensorTrait::new(
array![3, 3].span(),
array![
FP64x64 { mag: 18446744073709551616, sign: true },
FP64x64 { mag: 14757395258967642112, sign: true },
FP64x64 { mag: 11068046444225730560, sign: true },
FP64x64 { mag: 7378697629483821056, sign: true },
FP64x64 { mag: 3689348814741910528, sign: true },
FP64x64 { mag: 0, sign: false },
FP64x64 { mag: 3689348814741910528, sign: false },
FP64x64 { mag: 7378697629483821056, sign: false },
FP64x64 { mag: 11068046444225730560, sign: false }
]
.span()
);
return SVMClassifierTrait::predict(ref classifier, X);
}
>>> ([0, 1, 1],
[[0.728411, 0.271589],
[0.484705, 0.515295],
[0.274879, 0.725121]])Last updated