Merge pull request #311 from ELanning/fix-random-network-baseline

Update Examples from deprecated Pipeline to EnvironmentPipeline

Author: Hananel-Hazan

bindsnet/pipeline/action.py

@@ -26,7 +26,7 @@ def select_multinomial(pipeline: EnvironmentPipeline, **kwargs) -> int:
 
     assert (
         output.n % action_space.n == 0
-    ), "Output layer size not divisible by size of action space."
+    ), f"Output layer size of {output.n} is not divisible by action space size of {action_space.n}."
 
     pop_size = int(output.n / action_space.n)
     spikes = output.s
@@ -66,11 +66,11 @@ def select_softmax(pipeline: EnvironmentPipeline, **kwargs) -> int:
 
     assert (
         pipeline.network.layers[output].n == pipeline.env.action_space.n
-    ), "Output layer size not equal to size of action space."
+    ), "Output layer size is not equal to the size of the action space."
 
     assert hasattr(
         pipeline, "spike_record"
-    ), "EnvironmentPipeline has not attribute named: spike_record."
+    ), "EnvironmentPipeline is missing the attribute: spike_record."
 
     # Sum of previous iterations' spikes (Not yet implemented)
     spikes = torch.sum(pipeline.spike_record[output], dim=1)

examples/breakout/breakout.py

@@ -1,7 +1,5 @@
-import torch
-
 from bindsnet.network import Network
-from bindsnet.pipeline import Pipeline
+from bindsnet.pipeline import EnvironmentPipeline
 from bindsnet.encoding import bernoulli
 from bindsnet.network.topology import Connection
 from bindsnet.environment import GymEnvironment
@@ -27,12 +25,12 @@
 network.add_connection(inpt_middle, source="Input Layer", target="Hidden Layer")
 network.add_connection(middle_out, source="Hidden Layer", target="Output Layer")
 
-# Load SpaceInvaders environment.
+# Load the Breakout environment.
 environment = GymEnvironment("BreakoutDeterministic-v4")
 environment.reset()
 
 # Build pipeline from specified components.
-pipeline = Pipeline(
+pipeline = EnvironmentPipeline(
     network,
     environment,
     encoding=bernoulli,
@@ -47,12 +45,16 @@
 
 # Run environment simulation for 100 episodes.
 for i in range(100):
-    # initialize episode reward
-    reward = 0
+    total_reward = 0
     pipeline.reset_()
-    while True:
-        pipeline.step()
-        reward += pipeline.reward
-        if pipeline.done:
-            break
-    print("Episode " + str(i) + " reward:", reward)
+    is_done = False
+    while not is_done:
+        # Broken until select_softmax is implemented.
+        result = pipeline.env_step()
+        pipeline.step(result)
+
+        reward = result[1]
+        total_reward += reward
+
+        is_done = result[2]
+    print(f"Episode {i} total reward:{total_reward}")

examples/breakout/breakout_stdp.py

@@ -1,7 +1,5 @@
-import torch
-
 from bindsnet.network import Network
-from bindsnet.pipeline import Pipeline
+from bindsnet.pipeline import EnvironmentPipeline
 from bindsnet.learning import MSTDP
 from bindsnet.encoding import bernoulli
 from bindsnet.network.topology import Connection
@@ -36,12 +34,12 @@
 network.add_connection(inpt_middle, source="Input Layer", target="Hidden Layer")
 network.add_connection(middle_out, source="Hidden Layer", target="Output Layer")
 
-# Load SpaceInvaders environment.
+# Load the Breakout environment.
 environment = GymEnvironment("BreakoutDeterministic-v4")
 environment.reset()
 
 # Build pipeline from specified components.
-pipeline = Pipeline(
+environment_pipeline = EnvironmentPipeline(
     network,
     environment,
     encoding=bernoulli,
@@ -55,30 +53,28 @@
 )
 
 
-# Train agent for 100 episodes.
+def run_pipeline(pipeline, episode_count):
+    for i in range(episode_count):
+        total_reward = 0
+        pipeline.reset_()
+        is_done = False
+        while not is_done:
+            # Broken until select_softmax is implemented.
+            result = pipeline.env_step()
+            pipeline.step(result)
+
+            reward = result[1]
+            total_reward += reward
+
+            is_done = result[2]
+        print(f"Episode {i} total reward:{total_reward}")
+
+
 print("Training: ")
-for i in range(100):
-    pipeline.reset_()
-    # initialize episode reward
-    reward = 0
-    while True:
-        pipeline.step()
-        reward += pipeline.reward
-        if pipeline.done:
-            break
-    print("Episode " + str(i) + " reward:", reward)
+run_pipeline(environment_pipeline, episode_count=100)
 
 # stop MSTDP
-pipeline.network.learning = False
+environment_pipeline.network.learning = False
 
 print("Testing: ")
-for i in range(100):
-    pipeline.reset_()
-    # initialize episode reward
-    reward = 0
-    while True:
-        pipeline.step()
-        reward += pipeline.reward
-        if pipeline.done:
-            break
-    print("Episode " + str(i) + " reward:", reward)
+run_pipeline(environment_pipeline, episode_count=100)

examples/asteroids/random_network_baseline.py renamed to examples/breakout/random_network_baseline.py

@@ -3,7 +3,7 @@
 
 from bindsnet.network import Network
 from bindsnet.learning import Hebbian
-from bindsnet.pipeline import Pipeline
+from bindsnet.pipeline import EnvironmentPipeline
 from bindsnet.encoding import bernoulli
 from bindsnet.network.monitors import Monitor
 from bindsnet.environment import GymEnvironment
@@ -43,9 +43,9 @@
 network = Network(dt=dt)
 
 # Layers of neurons.
-inpt = Input(shape=(110, 84), traces=True)  # Input layer
+inpt = Input(shape=(80, 80), traces=True)  # Input layer
 exc = LIFNodes(n=n_neurons, refrac=0, traces=True)  # Excitatory layer
-readout = LIFNodes(n=14, refrac=0, traces=True)  # Readout layer
+readout = LIFNodes(n=16, refrac=0, traces=True)  # Readout layer
 layers = {"X": inpt, "E": exc, "R": readout}
 
 # Connections between layers.
@@ -93,11 +93,11 @@
     if layer in voltages:
         network.add_monitor(voltages[layer], name="%s_voltages" % layer)
 
-# Load SpaceInvaders environment.
-environment = GymEnvironment("Asteroids-v0")
+# Load the Breakout environment.
+environment = GymEnvironment("BreakoutDeterministic-v4")
 environment.reset()
 
-pipeline = Pipeline(
+pipeline = EnvironmentPipeline(
     network,
     environment,
     encoding=bernoulli,
@@ -120,9 +120,11 @@
 i = 0
 try:
     while i < n:
-        pipeline.step()
+        result = pipeline.env_step()
+        pipeline.step(result)
 
-        if pipeline.done:
+        is_done = result[2]
+        if is_done:
             pipeline.reset_()
 
         i += 1

examples/mnist/minimal_reservoir.py

@@ -9,7 +9,6 @@
 
 from bindsnet.network import Network
 from bindsnet.pipeline import DataLoaderPipeline
-from bindsnet.network.monitors import Monitor
 from bindsnet.network.topology import Connection
 from bindsnet.network.nodes import LIFNodes, Input
 

examples/space_invaders/et_space_invaders.py

@@ -11,7 +11,6 @@
 from bindsnet.network.topology import Connection
 from bindsnet.pipeline import EnvironmentPipeline
 from bindsnet.pipeline.action import select_multinomial
-from bindsnet.analysis.plotting import plot_weights
 
 
 parser = argparse.ArgumentParser()