oikan 0.0.3.2__py3-none-any.whl → 0.0.3.3__py3-none-any.whl

oikan/exceptions.py

@@ -4,4 +4,28 @@ class OIKANError(Exception):
 
 class ModelNotFittedError(OIKANError):
     """Raised when a method requires a fitted model."""
-    pass
+    pass
+
+class InvalidParameterError(OIKANError):
+    """Raised when an invalid parameter value is provided."""
+    pass
+
+class DataDimensionError(OIKANError):
+    """Raised when input data has incorrect dimensions."""
+    pass
+
+class NumericalInstabilityError(OIKANError):
+    """Raised when numerical computations become unstable."""
+    pass
+
+class FeatureExtractionError(OIKANError):
+    """Raised when feature extraction or transformation fails."""
+    pass
+
+class ModelSerializationError(OIKANError):
+    """Raised when model saving/loading operations fail."""
+    pass
+
+class ConvergenceError(OIKANError):
+    """Raised when the model fails to converge during training."""
+    pass    

oikan/model.py

@@ -3,13 +3,14 @@ import torch
 import torch.nn as nn
 import torch.optim as optim
 from sklearn.preprocessing import PolynomialFeatures
-from sklearn.linear_model import Lasso
+from sklearn.linear_model import ElasticNet
 from abc import ABC, abstractmethod
 import json
 from .neural import TabularNet
 from .utils import evaluate_basis_functions, get_features_involved
 from sklearn.model_selection import train_test_split
 from sklearn.metrics import r2_score, accuracy_score
+from .exceptions import *
 import sys
 
 class OIKAN(ABC):
@@ -30,6 +31,8 @@ class OIKAN(ABC):
         L1 regularization strength for Lasso in symbolic regression.
     sigma : float, optional (default=0.1)
         Standard deviation of Gaussian noise for data augmentation.
+    top_k : int, optional (default=5)
+        Number of top features to select in hierarchical symbolic regression.
     epochs : int, optional (default=100)
         Number of epochs for neural network training.
     lr : float, optional (default=0.001)
@@ -43,7 +46,28 @@ class OIKAN(ABC):
     """
     def __init__(self, hidden_sizes=[64, 64], activation='relu', augmentation_factor=10, 
                  polynomial_degree=2, alpha=0.1, sigma=0.1, epochs=100, lr=0.001, batch_size=32, 
-                 verbose=False, evaluate_nn=False):
+                 verbose=False, evaluate_nn=False, top_k=5):
+        if not isinstance(hidden_sizes, list) or not all(isinstance(x, int) and x > 0 for x in hidden_sizes):
+            raise InvalidParameterError("hidden_sizes must be a list of positive integers")
+        if activation not in ['relu', 'tanh', 'leaky_relu', 'elu', 'swish', 'gelu']:
+            raise InvalidParameterError(f"Unsupported activation function: {activation}")
+        if not isinstance(augmentation_factor, int) or augmentation_factor < 1:
+            raise InvalidParameterError("augmentation_factor must be a positive integer")
+        if not isinstance(polynomial_degree, int) or polynomial_degree < 1:
+            raise InvalidParameterError("polynomial_degree must be a positive integer")
+        if not isinstance(top_k, int) or top_k < 1:
+            raise InvalidParameterError("top_k must be a positive integer")
+        if not 0 < lr < 1:
+            raise InvalidParameterError("Learning rate must be between 0 and 1")
+        if not isinstance(batch_size, int) or batch_size < 1:
+            raise InvalidParameterError("batch_size must be a positive integer")
+        if not isinstance(epochs, int) or epochs < 1:
+            raise InvalidParameterError("epochs must be a positive integer")
+        if not 0 <= alpha <= 1:
+            raise InvalidParameterError("alpha must be between 0 and 1")
+        if sigma <= 0:
+            raise InvalidParameterError("sigma must be positive")
+        
         self.hidden_sizes = hidden_sizes
         self.activation = activation
         self.augmentation_factor = augmentation_factor
@@ -55,6 +79,7 @@ class OIKAN(ABC):
         self.batch_size = batch_size
         self.verbose = verbose
         self.evaluate_nn = evaluate_nn
+        self.top_k = top_k
         self.neural_net = None
         self.symbolic_model = None
         self.evaluation_done = False
@@ -74,13 +99,13 @@ class OIKAN(ABC):
         basis_functions = self.symbolic_model['basis_functions']
         if 'coefficients' in self.symbolic_model:
             coefficients = self.symbolic_model['coefficients']
-            formula = " + ".join([f"{coefficients[i]:.3f}*{basis_functions[i]}" 
+            formula = " + ".join([f"{coefficients[i]:.5f}*{basis_functions[i]}" 
                                 for i in range(len(coefficients)) if coefficients[i] != 0])
             return formula if formula else "0"
         else:
             formulas = []
             for c, coef in enumerate(self.symbolic_model['coefficients_list']):
-                formula = " + ".join([f"{coef[i]:.3f}*{basis_functions[i]}" 
+                formula = " + ".join([f"{coef[i]:.5f}*{basis_functions[i]}" 
                                     for i in range(len(coef)) if coef[i] != 0])
                 formulas.append(f"Class {self.classes_[c]}: {formula if formula else '0'}")
             return formulas
@@ -129,27 +154,33 @@ class OIKAN(ABC):
             File path to save the model. Should end with .json
         """
         if self.symbolic_model is None:
-            raise ValueError("Model not fitted yet.")
-            
+            raise ModelNotFittedError("Model must be fitted before saving")
+        
         if not path.endswith('.json'):
             path = path + '.json'
-            
-        # Convert numpy arrays and other non-serializable types to lists
-        model_data = {
-            'n_features': self.symbolic_model['n_features'],
-            'degree': self.symbolic_model['degree'],
-            'basis_functions': self.symbolic_model['basis_functions']
-        }
         
-        if 'coefficients' in self.symbolic_model:
-            model_data['coefficients'] = self.symbolic_model['coefficients']
-        else:
-            model_data['coefficients_list'] = [coef for coef in self.symbolic_model['coefficients_list']]
-            if hasattr(self, 'classes_'):
-                model_data['classes'] = self.classes_.tolist()
+        try:    
+            # Convert numpy arrays and other non-serializable types to lists
+            model_data = {
+                'n_features': self.symbolic_model['n_features'],
+                'degree': self.symbolic_model['degree'],
+                'basis_functions': self.symbolic_model['basis_functions']
+            }
+            
+            if 'coefficients' in self.symbolic_model:
+                model_data['coefficients'] = self.symbolic_model['coefficients']
+            else:
+                model_data['coefficients_list'] = [coef for coef in self.symbolic_model['coefficients_list']]
+                if hasattr(self, 'classes_'):
+                    model_data['classes'] = self.classes_.tolist()
+            
+            with open(path, 'w') as f:
+                json.dump(model_data, f, indent=2)
+        except Exception as e:
+            raise ModelSerializationError(f"Failed to save model: {str(e)}")
         
-        with open(path, 'w') as f:
-            json.dump(model_data, f, indent=2)
+        if self.verbose:
+            print(f"Model saved to {path}")
 
     def load(self, path):
         """
@@ -162,22 +193,28 @@ class OIKAN(ABC):
         """
         if not path.endswith('.json'):
             path = path + '.json'
+        
+        try:
+            with open(path, 'r') as f:
+                model_data = json.load(f)
+                
+            self.symbolic_model = {
+                'n_features': model_data['n_features'],
+                'degree': model_data['degree'],
+                'basis_functions': model_data['basis_functions']
+            }
             
-        with open(path, 'r') as f:
-            model_data = json.load(f)
-            
-        self.symbolic_model = {
-            'n_features': model_data['n_features'],
-            'degree': model_data['degree'],
-            'basis_functions': model_data['basis_functions']
-        }
+            if 'coefficients' in model_data:
+                self.symbolic_model['coefficients'] = model_data['coefficients']
+            else:
+                self.symbolic_model['coefficients_list'] = model_data['coefficients_list']
+                if 'classes' in model_data:
+                    self.classes_ = np.array(model_data['classes'])
+        except Exception as e:
+            raise ModelSerializationError(f"Failed to load model: {str(e)}")
         
-        if 'coefficients' in model_data:
-            self.symbolic_model['coefficients'] = model_data['coefficients']
-        else:
-            self.symbolic_model['coefficients_list'] = model_data['coefficients_list']
-            if 'classes' in model_data:
-                self.classes_ = np.array(model_data['classes'])
+        if self.verbose:
+            print(f"Model loaded from {path}")
 
     def _evaluate_neural_net(self, X, y, output_size, loss_fn):
         """Evaluates neural network performance on train-test split."""
@@ -253,7 +290,6 @@ class OIKAN(ABC):
 
     def _generate_augmented_data(self, X):
         """Generates augmented data by adding Gaussian noise."""
-        n_samples = X.shape[0]
         X_aug = []
         for _ in range(self.augmentation_factor):
             noise = np.random.normal(0, self.sigma, X.shape)
@@ -262,32 +298,102 @@ class OIKAN(ABC):
         return np.vstack(X_aug)
 
     def _perform_symbolic_regression(self, X, y):
-        """Performs symbolic regression using polynomial features and Lasso."""
-        poly = PolynomialFeatures(degree=self.polynomial_degree, include_bias=True)
-        X_poly = poly.fit_transform(X)
-        model = Lasso(alpha=self.alpha, fit_intercept=False)
-        model.fit(X_poly, y)
+        """
+        Performs hierarchical symbolic regression using a two-stage approach.
+        
+        Parameters:
+        -----------
+        X : array-like of shape (n_samples, n_features)
+            Input data.
+        y : array-like of shape (n_samples,) or (n_samples, n_classes)
+            Target values or logits.
+        """
+        n_features = X.shape[1]
+        self.top_k = min(self.top_k, n_features)
+        
+        if self.top_k < 1:
+            raise InvalidParameterError("top_k must be at least 1")
+
+        if np.any(np.isnan(X)) or np.any(np.isnan(y)):
+            raise NumericalInstabilityError("Input data contains NaN values")
+
+        if np.any(np.isinf(X)) or np.any(np.isinf(y)):
+            raise NumericalInstabilityError("Input data contains infinite values")
+
+        # Stage 1: Coarse Model
+        coarse_degree = 2  # Fixed low degree for coarse model
+        poly_coarse = PolynomialFeatures(degree=coarse_degree, include_bias=True)
+        X_poly_coarse = poly_coarse.fit_transform(X)
+        model_coarse = ElasticNet(alpha=self.alpha, fit_intercept=False)
+        model_coarse.fit(X_poly_coarse, y)
+
+        # Compute feature importances for original features
+        basis_functions_coarse = poly_coarse.get_feature_names_out()
         if len(y.shape) == 1 or y.shape[1] == 1:
-            coef = model.coef_.flatten()
-            selected_indices = np.where(np.abs(coef) > 1e-6)[0]
+            coef_coarse = model_coarse.coef_.flatten()
+        else:
+            coef_coarse = np.sum(np.abs(model_coarse.coef_), axis=0)
+
+        importances = np.zeros(X.shape[1])
+        for i, func in enumerate(basis_functions_coarse):
+            features_involved = get_features_involved(func)
+            for idx in features_involved:
+                importances[idx] += np.abs(coef_coarse[i])
+        
+        if np.all(importances == 0):
+            raise FeatureExtractionError("Failed to compute feature importances - all values are zero")
+
+        # Select top K features
+        top_k_indices = np.argsort(importances)[::-1][:self.top_k]
+
+        # Stage 2: Refined Model
+        # ~ generate additional non-linear features for top K features
+        additional_features = []
+        additional_names = []
+        for i in top_k_indices:
+            # Higher-degree polynomial
+            additional_features.append(X[:, i]**3)
+            additional_names.append(f'x{i}^3')
+            # Non-linear transformations
+            additional_features.append(np.log1p(np.abs(X[:, i])))
+            additional_names.append(f'log1p_x{i}')
+            additional_features.append(np.exp(np.clip(X[:, i], -10, 10)))
+            additional_names.append(f'exp_x{i}')
+            additional_features.append(np.sin(X[:, i]))
+            additional_names.append(f'sin_x{i}')
+        
+        # Combine features
+        X_additional = np.column_stack(additional_features)
+        X_refined = np.hstack([X_poly_coarse, X_additional])
+        basis_functions_refined = list(basis_functions_coarse) + additional_names
+
+        # Fit refined model
+        model_refined = ElasticNet(alpha=self.alpha, fit_intercept=False)
+        model_refined.fit(X_refined, y)
+
+        # Store symbolic model
+        if len(y.shape) == 1 or y.shape[1] == 1:
+            # Regression
+            coef_refined = model_refined.coef_.flatten()
+            selected_indices = np.where(np.abs(coef_refined) > 1e-6)[0]
             self.symbolic_model = {
                 'n_features': X.shape[1],
-                'degree': self.polynomial_degree,
-                'basis_functions': poly.get_feature_names_out()[selected_indices].tolist(),
-                'coefficients': coef[selected_indices].tolist()
+                'degree': self.polynomial_degree, 
+                'basis_functions': [basis_functions_refined[i] for i in selected_indices],
+                'coefficients': coef_refined[selected_indices].tolist()
             }
         else:
+            # Classification
             coefficients_list = []
-            # Note: Using the same basis functions across classes for simplicity
             selected_indices = set()
             for c in range(y.shape[1]):
-                coef = model.coef_[c]
+                coef = model_refined.coef_[c]
                 indices = np.where(np.abs(coef) > 1e-6)[0]
                 selected_indices.update(indices)
             selected_indices = list(selected_indices)
-            basis_functions = poly.get_feature_names_out()[selected_indices].tolist()
+            basis_functions = [basis_functions_refined[i] for i in selected_indices]
             for c in range(y.shape[1]):
-                coef = model.coef_[c]
+                coef = model_refined.coef_[c]
                 coef_selected = coef[selected_indices].tolist()
                 coefficients_list.append(coef_selected)
             self.symbolic_model = {

oikan/utils.py

@@ -9,7 +9,7 @@ def evaluate_basis_functions(X, basis_functions, n_features):
     X : array-like of shape (n_samples, n_features)
         Input data.
     basis_functions : list
-        List of basis function strings (e.g., '1', 'x0', 'x0^2', 'x0 x1').
+        List of basis function strings (e.g., '1', 'x0', 'x0^2', 'x0 x1', 'log1p_x0').
     n_features : int
         Number of input features.
     
@@ -22,15 +22,26 @@ def evaluate_basis_functions(X, basis_functions, n_features):
     for i, func in enumerate(basis_functions):
         if func == '1':
             X_transformed[:, i] = 1
+        elif func.startswith('log1p_x'):
+            idx = int(func.split('_')[1][1:])
+            X_transformed[:, i] = np.log1p(np.abs(X[:, idx]))
+        elif func.startswith('exp_x'):
+            idx = int(func.split('_')[1][1:])
+            X_transformed[:, i] = np.exp(np.clip(X[:, idx], -10, 10))
+        elif func.startswith('sin_x'):
+            idx = int(func.split('_')[1][1:])
+            X_transformed[:, i] = np.sin(X[:, idx])
         elif '^' in func:
             var, power = func.split('^')
             idx = int(var[1:])
             X_transformed[:, i] = X[:, idx] ** int(power)
         elif ' ' in func:
-            var1, var2 = func.split(' ')
-            idx1 = int(var1[1:])
-            idx2 = int(var2[1:])
-            X_transformed[:, i] = X[:, idx1] * X[:, idx2]
+            vars = func.split(' ')
+            result = np.ones(X.shape[0])
+            for var in vars:
+                idx = int(var[1:])
+                result *= X[:, idx]
+            X_transformed[:, i] = result
         else:
             idx = int(func[1:])
             X_transformed[:, i] = X[:, idx]
@@ -43,21 +54,29 @@ def get_features_involved(basis_function):
     Parameters:
     -----------
     basis_function : str
-        String representation of the basis function, e.g., 'x0', 'x0^2', 'x0 x1'.
+        String representation of the basis function, e.g., 'x0', 'x0^2', 'x0 x1', 'log1p_x0'.
     
     Returns:
     --------
     set : Set of feature indices involved.
     """
-    if basis_function == '1':  # Constant term involves no features
+    if basis_function == '1':
         return set()
     features = set()
-    for part in basis_function.split():  # Split by space for interaction terms
-        if part.startswith('x'):
-            if '^' in part:  # Handle powers, e.g., 'x0^2'
-                var = part.split('^')[0]  # Take 'x0'
-            else:
-                var = part  # Take 'x0' as is
-            idx = int(var[1:])  # Extract index, e.g., 0
+    if '_' in basis_function:  # Handle non-linear functions like 'log1p_x0'
+        parts = basis_function.split('_')
+        if len(parts) == 2 and parts[1].startswith('x'):
+            idx = int(parts[1][1:])
             features.add(idx)
+    elif '^' in basis_function:  # Handle powers, e.g., 'x0^2'
+        var = basis_function.split('^')[0]
+        idx = int(var[1:])
+        features.add(idx)
+    elif ' ' in basis_function:  # Handle interactions, e.g., 'x0 x1'
+        for part in basis_function.split():
+            idx = int(part[1:])
+            features.add(idx)
+    elif basis_function.startswith('x'):
+        idx = int(basis_function[1:])
+        features.add(idx)
     return features

{oikan-0.0.3.2.dist-info → oikan-0.0.3.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: oikan
-Version: 0.0.3.2
+Version: 0.0.3.3
 Summary: OIKAN: Neuro-Symbolic ML for Scientific Discovery
 Author: Arman Zhalgasbayev
 License: MIT
@@ -57,7 +57,7 @@ OIKAN implements a modern interpretation of the Kolmogorov-Arnold Representation
 
 2. **Neural Implementation**: OIKAN uses a specialized architecture combining:
    - Feature transformation layers with interpretable basis functions
-   - Symbolic regression for formula extraction
+   - Symbolic regression for formula extraction (ElasticNet-based)
    - Automatic pruning of insignificant terms
    
    ```python
@@ -76,15 +76,19 @@ OIKAN implements a modern interpretation of the Kolmogorov-Arnold Representation
    SYMBOLIC_FUNCTIONS = {
        'linear': 'x',           # Direct relationships
        'quadratic': 'x^2',      # Non-linear patterns
+       'cubic': 'x^3',         # Higher-order relationships
        'interaction': 'x_i x_j', # Feature interactions
-       'higher_order': 'x^n'    # Polynomial terms
+       'higher_order': 'x^n',    # Polynomial terms
+       'trigonometric': 'sin(x)', # Trigonometric functions
+       'exponential': 'exp(x)',  # Exponential growth
+       'logarithmic': 'log(x)'  # Logarithmic relationships
    }
    ```
 
 4. **Formula Extraction Process**:
    - Train neural network on raw data
    - Generate augmented samples for better coverage
-   - Perform L1-regularized symbolic regression
+   - Perform L1-regularized symbolic regression (alpha)
    - Prune terms with coefficients below threshold
    - Export human-readable mathematical expressions
 
@@ -115,12 +119,14 @@ model = OIKANRegressor(
     activation='relu', # Activation function (other options: 'tanh', 'leaky_relu', 'elu', 'swish', 'gelu')
     augmentation_factor=5, # Augmentation factor for data generation
     polynomial_degree=2, # Degree of polynomial basis functions
-    alpha=0.1, # L1 regularization strength
+    alpha=0.1, # L1 regularization strength (Symbolic regression)
     sigma=0.1, # Standard deviation of Gaussian noise for data augmentation
+    top_k=5, # Number of top features to select (Symbolic regression)
     epochs=100, # Number of training epochs
     lr=0.001, # Learning rate
     batch_size=32, # Batch size for training
-    verbose=True # Verbose output during training
+    verbose=True, # Verbose output during training
+    evaluate_nn=True # Validate neural network performance before full process
 )
 
 # Fit the model
@@ -163,12 +169,14 @@ model = OIKANClassifier(
     activation='relu', # Activation function (other options: 'tanh', 'leaky_relu', 'elu', 'swish', 'gelu')
     augmentation_factor=10, # Augmentation factor for data generation
     polynomial_degree=2, # Degree of polynomial basis functions
-    alpha=0.1, # L1 regularization strength
+    alpha=0.1, # L1 regularization strength (Symbolic regression)
     sigma=0.1, # Standard deviation of Gaussian noise for data augmentation
+    top_k=5, # Number of top features to select (Symbolic regression)
     epochs=100, # # Number of training epochs
     lr=0.001, # Learning rate
     batch_size=32, # Batch size for training
-    verbose=True # Verbose output during training
+    verbose=True, # Verbose output during training
+    evaluate_nn=True # Validate neural network performance before full process
 )
 
 # Fit the model

oikan-0.0.3.3.dist-info/RECORD

@@ -0,0 +1,10 @@
+oikan/__init__.py,sha256=zEzhm1GYLT4vNaIQ4CgZcNpUk3uo8SWnoaHYtHW_XSQ,628
+oikan/exceptions.py,sha256=GhHWqy2Q5LVBcteTy4ngnqxr7FOoLNyD8dNt1kfRXyw,901
+oikan/model.py,sha256=wvF_g1RcpYcQin_wOUiWEUeKJcQ8HyPtEm_5YrCeXFs,21946
+oikan/neural.py,sha256=wxmGgzmtpwJ3lvH6u6D4i4BiAzg018czrIdw49phSCY,1558
+oikan/utils.py,sha256=_FNhB_sIQfY-KsKRqvuKKVXNVZaAdpI5w8zPY_j_xJU,2898
+oikan-0.0.3.3.dist-info/licenses/LICENSE,sha256=75ASVmU-XIpN-M4LbVmJ_ibgbzbvRLVti8FhnR0BTf8,1096
+oikan-0.0.3.3.dist-info/METADATA,sha256=moaO5H0kXU-Gf_sV7tpt4VUgmTEys6dINlzr0yfDSUc,9055
+oikan-0.0.3.3.dist-info/WHEEL,sha256=DnLRTWE75wApRYVsjgc6wsVswC54sMSJhAEd4xhDpBk,91
+oikan-0.0.3.3.dist-info/top_level.txt,sha256=XwnwKwTJddZwIvtrUsAz-l-58BJRj6HjAGWrfYi_3QY,6
+oikan-0.0.3.3.dist-info/RECORD,,

{oikan-0.0.3.2.dist-info → oikan-0.0.3.3.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (80.3.1)
+Generator: setuptools (80.4.0)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

oikan-0.0.3.2.dist-info/RECORD

@@ -1,10 +0,0 @@
-oikan/__init__.py,sha256=zEzhm1GYLT4vNaIQ4CgZcNpUk3uo8SWnoaHYtHW_XSQ,628
-oikan/exceptions.py,sha256=Is0jG4apxO8QJQREIiJQYMjANYWibWeS-103q9KWbfg,192
-oikan/model.py,sha256=oZtx5uFxMj4q89ODKDBeTJ0whbqiMIR2tKwgmYVXHUY,16887
-oikan/neural.py,sha256=wxmGgzmtpwJ3lvH6u6D4i4BiAzg018czrIdw49phSCY,1558
-oikan/utils.py,sha256=xMGRa1qhn8BWn9UxpVeJIuGb-UvQmbjiFSsvAdF0bMU,2095
-oikan-0.0.3.2.dist-info/licenses/LICENSE,sha256=75ASVmU-XIpN-M4LbVmJ_ibgbzbvRLVti8FhnR0BTf8,1096
-oikan-0.0.3.2.dist-info/METADATA,sha256=PPsMSll3Ds6E9J3ZnXxo8Yno0ZsHrSb55kZ-035jJZE,8441
-oikan-0.0.3.2.dist-info/WHEEL,sha256=0CuiUZ_p9E4cD6NyLD6UG80LBXYyiSYZOKDm5lp32xk,91
-oikan-0.0.3.2.dist-info/top_level.txt,sha256=XwnwKwTJddZwIvtrUsAz-l-58BJRj6HjAGWrfYi_3QY,6
-oikan-0.0.3.2.dist-info/RECORD,,