Outlined NeuralNet abstraction

nn/nn.go

@@ -1 +1,35 @@
 package nn
+
+import (
+ "fmt"
+ "sync"
+
+ "gitlab.com/deepcypher/darklantern/dager"
+)
+
+type neuralNet struct {
+ graph dager.Dager
+ // optimiser
+ // loss function
+ lock sync.RWMutex
+}
+
+type NeuralNet interface {
+ train() error
+ infer() error
+}
+
+// NewNeuralNet initialises a new neural network
+func NewNeuralNet() NeuralNet {
+ return &neuralNet{}
+}
+
+func (nn *neuralNet) train() error {
+ fmt.Println("Training")
+ return nil
+}
+
+func (nn *neuralNet) infer() error {
+ fmt.Println("Infering")
+ return nil
+}

nn/nn_test.go

@@ -1,8 +1,11 @@
 package nn
 
 import (
+ "fmt"
  "testing"
 )
 
 func TestNNInit(t *testing.T) {
+ nn := NewNeuralNet()
+ fmt.Println(nn)
 }