Update MSE and BinaryCrossEntropy.

hallvardnmbu · May 19, 2024 · ed5b132 · ed5b132
1 parent 1354d14
commit ed5b132
Show file tree

Hide file tree

Showing 2 changed files with 32 additions and 32 deletions.
diff --git a/examples/example_train.rs → examples/train.rs b/examples/example_train.rs → examples/train.rs
@@ -32,16 +32,16 @@ fn main() {
     let nodes = vec![2, 2, 1];
     let biases = vec![false, false];
     let activations = vec![Activation::Sigmoid, Activation::Sigmoid];
-    let lr = 0.9f32;
+    let lr = 9.0f32;
     let optimizer = Optimizer::SGD;
-    let objective = Objective::BinaryCrossEntropy;
+    let objective = Objective::MSE;
 
     let mut net = network::Network::create(
         nodes, biases, activations, lr, optimizer, objective
     );
 
     // Train the network
-    let epoch_loss = net.train(&inputs, &targets, 1000);
+    let _epoch_loss = net.train(&inputs, &targets, 1000);
 
     // Validate the network
     let val_loss = net.validate(&inputs, &targets);

diff --git a/src/objective.rs b/src/objective.rs
@@ -19,9 +19,9 @@ use std::fmt::Display;
 pub enum Objective {
     AE,
     MAE,
-    // MSE,
+    MSE,
     RMSE,
-    // BinaryCrossEntropy,
+    BinaryCrossEntropy,
     CategoricalCrossEntropy,
 }
 
@@ -34,9 +34,9 @@ impl Display for Function {
         match self.objective {
             Objective::AE => write!(f, "AE"),
             Objective::MAE => write!(f, "MAE"),
-            // Objective::MSE => write!(f, "MSE"),
+            Objective::MSE => write!(f, "MSE"),
             Objective::RMSE => write!(f, "RMSE"),
-            // Objective::BinaryCrossEntropy => write!(f, "BinaryCrossEntropy"),
+            Objective::BinaryCrossEntropy => write!(f, "BinaryCrossEntropy"),
             Objective::CategoricalCrossEntropy => write!(f, "CategoricalCrossEntropy"),
         }
     }
@@ -47,9 +47,9 @@ impl Function {
         match objective {
             Objective::AE => Function { objective: Objective::AE },
             Objective::MAE => Function { objective: Objective::MAE },
-            // Objective::MSE => Function { objective: Objective::MSE },
+            Objective::MSE => Function { objective: Objective::MSE },
             Objective::RMSE => Function { objective: Objective::RMSE },
-            // Objective::BinaryCrossEntropy => Function { objective: Objective::BinaryCrossEntropy },
+            Objective::BinaryCrossEntropy => Function { objective: Objective::BinaryCrossEntropy },
             Objective::CategoricalCrossEntropy => Function { objective: Objective::CategoricalCrossEntropy },
         }
     }
@@ -88,15 +88,15 @@ impl Function {
                     ).collect();
                 (loss, gradient)
             },
-            // Objective::MSE => {
-            //     let loss: f32 = y.iter().zip(out.iter())
-            //         .map(|(actual, predicted)| (actual - predicted).powi(2))
-            //         .sum::<f32>() / y.len() as f32;
-            //     let gradient: Vec<f32> = y.iter().zip(out.iter())
-            //         .map(|(actual, predicted)| 2.0 * (actual - predicted) / y.len() as f32)
-            //         .collect();
-            //     (loss, gradient)
-            // },
+            Objective::MSE => {
+                let loss: f32 = y.iter().zip(out.iter())
+                    .map(|(actual, predicted)| (actual - predicted).powi(2))
+                    .sum::<f32>() / y.len() as f32;
+                let gradient: Vec<f32> = y.iter().zip(out.iter())
+                    .map(|(actual, predicted)| -2.0 * (actual - predicted) / y.len() as f32)
+                    .collect();
+                (loss, gradient)
+            },
             Objective::RMSE => {
                 let loss: f32 = y.iter().zip(out.iter())
                     .map(|(actual, predicted)| (actual - predicted).powi(2))
@@ -112,20 +112,20 @@ impl Function {
                     ).collect();
                 (loss, gradient)
             },
-            // Objective::BinaryCrossEntropy => {
-            //     let eps: f32 = 1e-7;
-            //     let loss: f32 = -y.iter().zip(out.iter())
-            //         .map(|(actual, predicted)| {
-            //                 let predicted = predicted.clamp(eps, 1.0 - eps);
-            //                 actual * predicted.ln() + (1.0 - actual) * (1.0 - predicted).ln()
-            //         }).sum::<f32>() / y.len() as f32;
-            //     let gradient: Vec<f32> = y.iter().zip(out.iter())
-            //         .map(|(actual, predicted)| {
-            //             let predicted = predicted.clamp(eps, 1.0 - eps);
-            //             (predicted - actual) / (predicted * (1.0 - predicted))
-            //         }).collect();
-            //     (loss, gradient)
-            // },
+            Objective::BinaryCrossEntropy => {
+                let eps: f32 = 1e-7;
+                let loss: f32 = -y.iter().zip(out.iter())
+                    .map(|(actual, predicted)| {
+                            let predicted = predicted.clamp(eps, 1.0 - eps);
+                            actual * predicted.ln() + (1.0 - actual) * (1.0 - predicted).ln()
+                    }).sum::<f32>() / y.len() as f32;
+                let gradient: Vec<f32> = y.iter().zip(out.iter())
+                    .map(|(actual, predicted)| {
+                        let predicted = predicted.clamp(eps, 1.0 - eps);
+                        (predicted - actual) / (predicted * (1.0 - predicted))
+                    }).collect();
+                (loss, gradient)
+            },
             Objective::CategoricalCrossEntropy => {
                 let eps: f32 = 1e-7;
                 let loss: f32 = -y.iter().zip(out.iter())