artlib.optimized.backends.torch.BinaryFuzzyARTMAP

Binary Fuzzy ARTMAP (Torch-accelerated backend)

Classes

`_TorchBinaryFuzzyARTMAPConfig`
`_TorchBinaryFuzzyARTMAP`	Torch-accelerated Binary Fuzzy ARTMAP with export hooks for artlib sync.
`BinaryFuzzyARTMAP`	BinaryFuzzyARTMAP for Classification. optimized with torch.

Functions

`_to_device`(→ torch.Tensor)
`_complement_code_any`(→ torch.Tensor)	Complement-code that works for float or bool tensors.

Module Contents

artlib.optimized.backends.torch.BinaryFuzzyARTMAP._to_device(x: torch.Tensor | numpy.ndarray, device, dtype=torch.bool) → torch.Tensor

artlib.optimized.backends.torch.BinaryFuzzyARTMAP._complement_code_any(x: torch.Tensor) → torch.Tensor: Complement-code that works for float or bool tensors.

class artlib.optimized.backends.torch.BinaryFuzzyARTMAP._TorchBinaryFuzzyARTMAPConfig

input_dim: int

alpha: float = 1e-10

rho: float = 0.75

epsilon: float = 1e-07

match_tracking: bool = True

device: str = 'cuda'

dtype: torch.dtype = Ellipsis

clamp_inputs: bool = True

fallback_to_choice_on_fail: bool = True

class artlib.optimized.backends.torch.BinaryFuzzyARTMAP._TorchBinaryFuzzyARTMAP(cfg: _TorchBinaryFuzzyARTMAPConfig)

Torch-accelerated Binary Fuzzy ARTMAP with export hooks for artlib sync.

cfg

device

dtype

input_dim

code_dim

W: torch.Tensor | None = None

map_y: torch.Tensor | None = None

_lower_bounds: torch.Tensor | None = None

_upper_bounds: torch.Tensor | None = None

_prep_tol: float = 1e-06

property n_cat: int

_ensure_capacity()

set_data_bounds(lower: torch.Tensor | numpy.ndarray, upper: torch.Tensor | numpy.ndarray)

prepare_data(X: torch.Tensor | numpy.ndarray) → torch.Tensor: Optional helper: normalize -> complement-code -> binarize -> bool. If your inputs are already binary and complement-coded, skip this and pass them straight to partial_fit_and_export/predict_*.

_validate_prepared(X: torch.Tensor)

_free_mem_bytes() → int

_choice_and_match(I: torch.Tensor) → Tuple[torch.Tensor, torch.Tensor, torch.Tensor]

For a single prepared binary input I (bool), return:

T: [K] choice values m: [K] match values w_sum: [K] |w| (as float64)

_commit_new_category(I: torch.Tensor, y: int)

partial_fit_and_export(X_prepared: torch.Tensor | numpy.ndarray, y: torch.Tensor | numpy.ndarray, epsilon: float = 1e-10, match_tracking: Literal['MT+', 'MT-', 'MT0', 'MT1', 'MT~'] = 'MT+') → Tuple[numpy.ndarray, list[numpy.ndarray], numpy.ndarray]

Incremental training on already-prepared, binary, complement-coded inputs.

Returns:: (N,) chosen A-side category indices (int) weights_arrays: list of K arrays, each weight as uint8 cluster_labels_out: (K,) map from A categories to B labels (int)
Return type:: labels_a_out

predict_ab_prepared(X_prepared: torch.Tensor | numpy.ndarray) → Tuple[numpy.ndarray, numpy.ndarray]

class artlib.optimized.backends.torch.BinaryFuzzyARTMAP.BinaryFuzzyARTMAP(rho: float, alpha: float, input_dim: int | None = None, device: str = 'cuda')

Bases: artlib.optimized.backends.torch._TorchSimpleARTMAP._TorchSimpleARTMAP

BinaryFuzzyARTMAP for Classification. optimized with torch.

This module implements BinaryFuzzyARTMAP

BinaryFuzzyARTMAP is a non-modular classification model which has been highly optimized for run-time performance. Fit and predict functions are implemented in torch for efficient execution. This class acts as a wrapper for the underlying torch functions and to provide compatibility with the artlib style and usage. Functionally, BinaryFuzzyARTMAP behaves as a special case of SimpleARTMAP instantiated with BinaryFuzzyART.

_device = 'cuda'

_backend: _TorchBinaryFuzzyARTMAP | None = None

_declared_input_dim = None

_ensure_backend(X: numpy.ndarray): Initialize backend using prepared X to infer raw dimension when needed.