Crate sklears_neural 

Neural network implementations for the sklears machine learning library.

This crate provides implementations of neural network algorithms compatible with the scikit-learn API, including Multi-Layer Perceptron (MLP) for classification and regression tasks.

Examples

use sklears_neural::{MLPClassifier, Activation};
use sklears_core::traits::{Predict};
use scirs2_core::ndarray::Array2;

let x = Array2::from_shape_vec((4, 2), vec![
    0.0, 0.0,
    0.0, 1.0,
    1.0, 0.0,
    1.0, 1.0,
])?;
let y = vec![0, 1, 1, 0]; // XOR problem

let mlp = MLPClassifier::new()
    .hidden_layer_sizes(&[10, 5])
    .activation(Activation::Relu)
    .max_iter(1000)
    .learning_rate_init(0.01)
    .random_state(42);

let trained_mlp = mlp.fit(&x, &y)?;
let predictions = trained_mlp.predict(&x)?;

Re-exports

pub use activation::*;

pub use attention_rnn::*;

pub use autoencoder::*;

pub use checkpointing::*;

pub use config::*;

pub use conv_layers::*;

pub use curriculum::*;

pub use data_augmentation::*;

pub use distributed::*;

pub use experiment_tracking::*;

pub use gan::*;

pub use gpu::*;

pub use gradient_checking::*;

pub use interpretation::*;

pub use knowledge_distillation::*;

pub use layers::*;

pub use memory_leak_tests::*;

pub use mlp_classifier::*;

pub use mlp_regressor::*;

pub use model_selection::*;

pub use models::*;

pub use multi_task::*;

pub use neural_metrics::*;

pub use performance_testing::*;

pub use quantization::*;

pub use rbm::*;

pub use regularization::*;

pub use self_supervised::*;

pub use seq2seq::*;

pub use solvers::*;

pub use transfer_learning::*;

pub use transformer::*;

pub use utils::*;

pub use vae::*;

pub use validation::*;

pub use versioning::*;

pub use visualization::*;

pub use weight_init::*;

Modules

activation

Activation functions for neural networks.

attention_rnn

Attention-based Recurrent Neural Networks

autoencoder

Autoencoder implementation for unsupervised feature learning

checkpointing

Model checkpointing and state management for neural networks.

config

Configuration Management for Neural Networks

conv_layers

Convolutional layers for neural networks.

curriculum

Curriculum learning implementation for neural networks.

data_augmentation

Data augmentation utilities for neural network training.

distributed

experiment_tracking

Comprehensive experiment tracking system for neural network training.

gan

Generative Adversarial Networks (GANs) implementation

gpu

GPU acceleration module for neural network computations

gradient_checking

interpretation

Model interpretation and explainability tools for neural networks.

knowledge_distillation

Knowledge Distillation implementation for model compression

layers

Neural network layer implementations.

memory_leak_tests

Memory leak detection tests for neural network training and inference.

mlp_classifier

Multi-layer Perceptron (MLP) classifier implementation.

mlp_regressor

Multi-layer Perceptron (MLP) regressor implementation.

model_selection

models

Neural network model architectures.

multi_task

Multi-Task Learning for Neural Networks

neural_metrics

Neural Network Specific Metrics

performance_testing

Performance testing and benchmarking utilities for neural networks.

quantization

Model Quantization utilities for neural network compression

rbm

Restricted Boltzmann Machine (RBM) implementation

regularization

Regularization techniques for neural networks.

self_supervised

seq2seq

Sequence-to-sequence models for neural machine translation and similar tasks.

solvers

Optimization solvers for neural networks.

transfer_learning

Transfer Learning Utilities

transformer

Encoder-Decoder Transformer Architectures

utils

Neural network utility functions and helper modules.

vae

Variational Autoencoder (VAE) implementation for generative modeling

validation

Hyperparameter validation and configuration management for neural networks.

versioning

visualization

Model Visualization Utilities

weight_init

Weight initialization strategies for neural networks.

Type Aliases

NeuralResult

Result type for neural network operations