def plot_spikes_inf(self):
plot = nengo_plot.Time(self.sim.trange())
plot.add('Visual Buffer', spa.similarity(self.sim.data, self.vision),
overlays=[(0.06,'Lin Task'),(0.25, 'Test Stimulus')])
plot.add_spikes('', self.sim.data[self.visSpikes],
sample_by_variance=64, cluster=True)
plot.add('Memory - Context', spa.similarity(self.sim.data, self.context),
overlays=[(0.085,'Inferential Evaluation'),(0.14, 'Rule 1'),
(0.19, 'Rule 2'),(0.24, 'Rule 3'),
(0.29, 'Rule 4'),(0.35, 'Motor Routing')])
plot.add_spikes('', self.sim.data[self.conSpikes],
sample_by_variance=64, cluster=True)
plot.add('Applications', spa.similarity(self.sim.data, self.apps),
overlays=[(0.2, 'Feature 1'),
(0.25, 'Feature 2'),(0.3, 'Feature 3'),
(0.35, 'Feature 4')])
plot.add_spikes('', self.sim.data[self.appSpikes],
sample_by_variance=64, cluster=True)
plot.add('Coherence', self.sim.data[self.score],
overlays=[(0.35, 'Coherence Value')])
plot.add_spikes('', self.sim.data[self.scoSpikes], sample_by_variance=64)
plot.add('Motor Buffer', spa.similarity(self.sim.data, self.motor),
overlays=[(0.42,'Positive Judgment')])
plot.add_spikes('', self.sim.data[self.motSpikes],
sample_by_variance=64, cluster=True)
plot.save('inf_raster.png')
def plot(self):
plots = 6
fig = plt.figure(figsize=(20,10))
p1 = fig.add_subplot(plots,1,1)
p1.plot(self.sim.trange(),spa.similarity(self.sim.data,self.vision))
p1.set_title('Visual Buffer', fontsize='15')
p2 = fig.add_subplot(plots,1,2)
p2.plot(self.sim.trange(),spa.similarity(self.sim.data,self.context))
p2.set_title('Task Context', fontsize='15')
p3 = fig.add_subplot(plots,1,3)
p3.plot(self.sim.trange(),self.sim.data[self.memory])
p3.set_title('Memory', fontsize='15')
p4 = fig.add_subplot(plots,1,4)
p4.plot(self.sim.trange(), self.sim.data[self.score])
p4.set_title('Coherence Score', fontsize='15')
p5 = fig.add_subplot(plots,1,5)
p5.plot(self.sim.trange(),spa.similarity(self.sim.data,self.decision))
p5.set_title('Decision Buffer', fontsize='15')
p5.set_ylim([-0.5,2])
p6 = fig.add_subplot(plots,1,6)
p6.plot(self.sim.trange(), spa.similarity(self.sim.data, self.motor))
p6.set_title('Motor', fontsize='15')
p6.legend(self.motor.target.vocab.keys, fontsize=12,
loc='upper center', bbox_to_anchor=(0.5, -0.2), ncol=14)
fig.subplots_adjust(hspace=0.65)
fig.savefig(os.path.join('results', str(self.probe)))
plt.close(fig)
def plot_spikes_vis(self):
plot = nengo_plot.Time(self.sim.trange())
plot.add('Visual Buffer',
spa.similarity(self.sim.data, self.vision),
overlays=[(0.06, 'Posner Task'), (0.25, 'Test Stimulus')])
plot.add_spikes('',
self.sim.data[self.visSpikes],
sample_by_variance=64,
cluster=True)
plot.add('Memory - Context',
spa.similarity(self.sim.data, self.context),
overlays=[(0.085, 'Perceptual Evaluation'),
(0.14, 'Motor Routing')])
plot.add_spikes('',
self.sim.data[self.conSpikes],
sample_by_variance=64,
cluster=True)
plot.add('Memory - SP',
spa.similarity(self.sim.data, self.memory),
overlays=[(0.25, 'Semantic Pointer')])
plot.add_spikes('',
self.sim.data[self.memSpikes],
sample_by_variance=64,
cluster=True)
plot.add('Motor Buffer',
spa.similarity(self.sim.data, self.motor),
overlays=[(0.25, 'Category Label A')])
plot.add_spikes('',
self.sim.data[self.motSpikes],
sample_by_variance=64,
cluster=True)
plot.save('vis_raster.png')
def plot_spikes_inf(self):
plot = nengo_plot.Time(self.sim.trange())
plot.add('Visual Buffer',
spa.similarity(self.sim.data, self.vision),
overlays=[(0.06, 'Lin Task'), (0.25, 'Test Stimulus')])
plot.add_spikes('',
self.sim.data[self.visSpikes],
sample_by_variance=64,
cluster=True)
plot.add('Memory - Context',
spa.similarity(self.sim.data, self.context),
overlays=[(0.085, 'Inferential Evaluation'), (0.14, 'Rule 1'),
(0.19, 'Rule 2'), (0.24, 'Rule 3'),
(0.29, 'Rule 4'), (0.35, 'Motor Routing')])
plot.add_spikes('',
self.sim.data[self.conSpikes],
sample_by_variance=64,
cluster=True)
plot.add('Applications',
spa.similarity(self.sim.data, self.apps),
overlays=[(0.2, 'Feature 1'), (0.25, 'Feature 2'),
(0.3, 'Feature 3'), (0.35, 'Feature 4')])
plot.add_spikes('',
self.sim.data[self.appSpikes],
sample_by_variance=64,
cluster=True)
plot.add('Coherence',
self.sim.data[self.score],
overlays=[(0.35, 'Coherence Value')])
plot.add_spikes('',
self.sim.data[self.scoSpikes],
sample_by_variance=64)
plot.add('Motor Buffer',
spa.similarity(self.sim.data, self.motor),
overlays=[(0.42, 'Positive Judgment')])
plot.add_spikes('',
self.sim.data[self.motSpikes],
sample_by_variance=64,
cluster=True)
plot.save('inf_raster.png')
def output_func(t, x):
similarity = spa.similarity(x, out_vocab.vectors)
if np.any(similarity > 0.5):
global g_motion_out
g_motion_out = out_vocab.keys[np.argmax(similarity)]
else:
global g_motion_out
g_motion_out = ''
global g_simulator_alive
g_simulator_alive = time.time()
def plot(self):
plots = 6
fig = plt.figure(figsize=(20, 10))
p1 = fig.add_subplot(plots, 1, 1)
p1.plot(self.sim.trange(), spa.similarity(self.sim.data, self.vision))
p1.set_title('Visual Buffer', fontsize='15')
p2 = fig.add_subplot(plots, 1, 2)
p2.plot(self.sim.trange(), spa.similarity(self.sim.data, self.context))
p2.set_title('Task Context', fontsize='15')
p3 = fig.add_subplot(plots, 1, 3)
p3.plot(self.sim.trange(), self.sim.data[self.memory])
p3.set_title('Memory', fontsize='15')
p4 = fig.add_subplot(plots, 1, 4)
p4.plot(self.sim.trange(), self.sim.data[self.score])
p4.set_title('Coherence Score', fontsize='15')
p5 = fig.add_subplot(plots, 1, 5)
p5.plot(self.sim.trange(), spa.similarity(self.sim.data,
self.decision))
p5.set_title('Decision Buffer', fontsize='15')
p5.set_ylim([-0.5, 2])
p6 = fig.add_subplot(plots, 1, 6)
p6.plot(self.sim.trange(), spa.similarity(self.sim.data, self.motor))
p6.set_title('Motor', fontsize='15')
p6.legend(self.motor.target.vocab.keys,
fontsize=12,
loc='upper center',
bbox_to_anchor=(0.5, -0.2),
ncol=14)
fig.subplots_adjust(hspace=0.65)
fig.savefig(os.path.join('results', str(self.probe)))
plt.close(fig)
def plot_spikes_vis(self):
plot = nengo_plot.Time(self.sim.trange())
plot.add('Visual Buffer', spa.similarity(self.sim.data, self.vision),
overlays=[(0.06,'Posner Task'),(0.25, 'Test Stimulus')])
plot.add_spikes('', self.sim.data[self.visSpikes],
sample_by_variance=64, cluster=True)
plot.add('Memory - Context', spa.similarity(self.sim.data, self.context),
overlays=[(0.085,'Perceptual Evaluation'),
(0.14,'Motor Routing')])
plot.add_spikes('', self.sim.data[self.conSpikes],
sample_by_variance=64, cluster=True)
plot.add('Memory - SP', spa.similarity(self.sim.data, self.memory),
overlays=[(0.25,'Semantic Pointer')])
plot.add_spikes('', self.sim.data[self.memSpikes],
sample_by_variance=64, cluster=True)
plot.add('Motor Buffer', spa.similarity(self.sim.data, self.motor),
overlays=[(0.25,'Category Label A')])
plot.add_spikes('', self.sim.data[self.motSpikes],
sample_by_variance=64, cluster=True)
plot.save('vis_raster.png')
def run_model(self):
res = RecognitionResult()
rng = np.random.RandomState(self.seed)
paths, freqs = analysis.get_syllables(
self.n_syllables, self.minfreq, self.maxfreq, rng)
for path, freq in zip(paths, freqs):
traj = vtl.parse_ges(path).trajectory(self.model.trial.dt)
self.model.add_syllable(label=path2label(path),
freq=freq,
trajectory=traj)
# Determine and save syllable sequence
if self.model.trial.repeat:
seq_ix = rng.randint(len(paths), size=self.sequence_len)
else:
seq_ix = rng.permutation(len(paths))[:self.sequence_len]
seq = [path2label(paths[i]) for i in seq_ix]
res.seq = np.array(seq)
# Determine how long to run
simt = 0.0
tgt_time = []
for label in res.seq:
syllable = self.model.syllable_dict[label]
simt += 1. / syllable.freq
tgt_time.append(simt)
# Set that sequence in the model
traj = ideal_traj(self.model, seq)
self.model.trial.trajectory = traj
# Save frequencies for that sequence
res.freqs = np.array([self.model.syllables[i].freq for i in seq_ix])
# -- Run the model
net = self.model.build(nengo.Network(seed=self.seed))
with net:
p_dmps = [nengo.Probe(dmp.state[0], synapse=0.01)
for dmp in net.syllables]
p_class = nengo.Probe(net.classifier, synapse=0.01)
p_mem = nengo.Probe(net.memory.output, synapse=0.01)
sim = nengo.Simulator(net)
sim.run(simt)
# Save iDMP system states
res.dmps = np.hstack([sim.data[p_d] for p_d in p_dmps])
res.dmp_labels = np.array([s.label for s in self.model.syllables])
# Save working memory similarities
res.memory = spa.similarity(sim.data[p_mem], net.vocab, True)
# Determine classification times and labels
t_ix, class_ix = analysis.classinfo(sim.data[p_class], res.dmps)
res.class_time = sim.trange()[t_ix]
res.class_labels = np.array([path2label(paths[ix]) for ix in class_ix])
# Calculate accuracy / timing metrics
recinfo = [(t, l) for t, l in zip(res.class_time, res.class_labels)]
tgtinfo = [(t, l) for t, l in zip(tgt_time, res.seq)]
res.acc, res.n_sub, res.n_del, res.n_ins = (
analysis.cl_accuracy(recinfo, tgtinfo))
res.tdiff_mean, res.tdiff_var = analysis.cl_timing(recinfo, tgtinfo)
log("Accuracy: %.3f" % res.acc)
# Determine if memory representation is correct
tgt_time = np.asarray(tgt_time)
mem_times = (tgt_time[1:] + tgt_time[:-1]) * 0.5
mem_ix = (mem_times / self.model.trial.dt).astype(int)
mem_class = np.argmax(res.memory[mem_ix], axis=1)
slabels = [s.label for s in self.model.syllables]
actual = np.array([slabels.index(lbl) for lbl in res.seq[:-1]])
res.memory_acc = np.mean(mem_class == actual)
return res
plt.subplot(5, 1, 1)
plt.plot(sim.trange(), model.similarity(sim.data, color_in))
plt.legend(model.get_output_vocab('color_in').keys, fontsize='x-small')
plt.ylabel("color")
plt.subplot(5, 1, 2)
plt.plot(sim.trange(), model.similarity(sim.data, shape_in))
plt.legend(model.get_output_vocab('shape_in').keys, fontsize='x-small')
plt.ylabel("shape")
plt.subplot(5, 1, 3)
plt.plot(sim.trange(), model.similarity(sim.data, cue))
plt.legend(model.get_output_vocab('cue').keys, fontsize='x-small')
plt.ylabel("cue")
plt.subplot(5, 1, 4)
for pointer in ['RED * CIRCLE', 'BLUE * SQUARE']:
plt.plot(sim.trange(), vocab.parse(pointer).dot(sim.data[conv].T), label=pointer)
plt.legend(fontsize='x-small')
plt.ylabel("convolved")
plt.subplot(5, 1, 5)
plt.plot(sim.trange(), spa.similarity(sim.data[out], vocab))
plt.legend(model.get_output_vocab('out').keys, fontsize='x-small')
plt.ylabel("output")
plt.xlabel("time [s]");
# The last plot shows that the output is most similar to the semantic pointer bound to the current cue. For example, when `RED` and `CIRCLE` are being convolved and the cue is `CIRCLE`, the output is most similar to `RED`.
sim = nengo.Simulator(model, dt=dt)
sim.run(T)
t = sim.trange()
if plot_res:
import matplotlib.pyplot as plt
# figure out how to put these into a subplot
plt.figure()
plt.title("Error")
plt.plot(t, np.linalg.norm(sim.data[p_error], axis=1))
plt.figure()
plt.title("Keys_1")
plt.plot(t, spa.similarity(sim.data[p_keys][:, :D], vocab))
plt.legend(vocab.keys, loc='best')
plt.figure()
plt.title("Keys_2")
plt.plot(t, spa.similarity(sim.data[p_keys][:, D:], vocab))
plt.legend(vocab.keys, loc='best')
plt.figure()
plt.title("Result")
plt.plot(t, spa.similarity(sim.data[p_recall], vocab))
plt.legend(vocab.keys, loc='best')
plt.ylim(-1.5, 1.5)
plt.figure()
plt.title("Actual Answer")
val_res[seed_val] = sim.data[p_values]
error_res[seed_val] = np.sum(np.abs(sim.data[p_error]), axis=1)
recall_res[seed_val] = sim.data[p_recall]
if plot_res:
import matplotlib.pyplot as plt
# figure out how to put these into a subplot
plt.figure()
plt.title("Error")
plt.plot(np.linalg.norm(sim.data[p_error], axis=1))
plt.figure()
plt.title("Result")
plt.plot(spa.similarity(sim.data[p_recall], vocab))
plt.legend(vocab.keys, loc='best')
plt.ylim(-0.5, 1.1)
plt.figure()
plt.title("Actual Answer")
plt.plot(spa.similarity(sim.data[p_values], vocab))
plt.ylim(-0.5, 1.1)
plt.show()
ipdb.set_trace()
# I should make a wrapper for doing this quickly
base_name = "multpred2"
np.savez_compressed("data/%s_learning_data" % base_name, p_keys=key_res, p_recall=recall_res, p_error=error_res, p_values=val_res)
np.savez_compressed("data/%s_learning_vocab" % base_name, keys=vocab.keys, vecs=vocab.vectors)
def run_model(self):
res = RecognitionResult()
rng = np.random.RandomState(self.seed)
paths, freqs = analysis.get_syllables(self.n_syllables, self.minfreq,
self.maxfreq, rng)
for path, freq in zip(paths, freqs):
traj = vtl.parse_ges(path).trajectory(self.model.trial.dt)
self.model.add_syllable(label=path2label(path),
freq=freq,
trajectory=traj)
# Determine and save syllable sequence
if self.model.trial.repeat:
seq_ix = rng.randint(len(paths), size=self.sequence_len)
else:
seq_ix = rng.permutation(len(paths))[:self.sequence_len]
seq = [path2label(paths[i]) for i in seq_ix]
res.seq = np.array(seq)
# Determine how long to run
simt = 0.0
tgt_time = []
for label in res.seq:
syllable = self.model.syllable_dict[label]
simt += 1. / syllable.freq
tgt_time.append(simt)
# Set that sequence in the model
traj = ideal_traj(self.model, seq)
self.model.trial.trajectory = traj
# Save frequencies for that sequence
res.freqs = np.array([self.model.syllables[i].freq for i in seq_ix])
# -- Run the model
net = self.model.build(nengo.Network(seed=self.seed))
with net:
p_dmps = [
nengo.Probe(dmp.state[0], synapse=0.01)
for dmp in net.syllables
]
p_class = nengo.Probe(net.classifier, synapse=0.01)
p_mem = nengo.Probe(net.memory.output, synapse=0.01)
sim = nengo.Simulator(net)
sim.run(simt)
# Save iDMP system states
res.dmps = np.hstack([sim.data[p_d] for p_d in p_dmps])
res.dmp_labels = np.array([s.label for s in self.model.syllables])
# Save working memory similarities
res.memory = spa.similarity(sim.data[p_mem], net.vocab, True)
# Determine classification times and labels
t_ix, class_ix = analysis.classinfo(sim.data[p_class], res.dmps)
res.class_time = sim.trange()[t_ix]
res.class_labels = np.array([path2label(paths[ix]) for ix in class_ix])
# Calculate accuracy / timing metrics
recinfo = [(t, l) for t, l in zip(res.class_time, res.class_labels)]
tgtinfo = [(t, l) for t, l in zip(tgt_time, res.seq)]
res.acc, res.n_sub, res.n_del, res.n_ins = (analysis.cl_accuracy(
recinfo, tgtinfo))
res.tdiff_mean, res.tdiff_var = analysis.cl_timing(recinfo, tgtinfo)
log("Accuracy: %.3f" % res.acc)
# Determine if memory representation is correct
tgt_time = np.asarray(tgt_time)
mem_times = (tgt_time[1:] + tgt_time[:-1]) * 0.5
mem_ix = (mem_times / self.model.trial.dt).astype(int)
mem_class = np.argmax(res.memory[mem_ix], axis=1)
slabels = [s.label for s in self.model.syllables]
actual = np.array([slabels.index(lbl) for lbl in res.seq[:-1]])
res.memory_acc = np.mean(mem_class == actual)
return res
# Calculate the error and use it to drive the PES rule
nengo.Connection(env_value, error, transform=-1, synapse=None)
nengo.Connection(recall, error, synapse=None)
nengo.Connection(error, het_mem.out_conn.learning_rule)
# Setup probes
p_keys = nengo.Probe(env_key, synapse=None, sample_every=sample_every)
p_values = nengo.Probe(env_value, synapse=None, sample_every=sample_every)
p_error = nengo.Probe(error, synapse=0.01)
p_out = nengo.Probe(het_mem.output, synapse=0.01, sample_every=sample_every)
p_recall = nengo.Probe(recall, synapse=None, sample_every=sample_every)
sim = nengo.Simulator(model, dt=dt)
sim.run(period)
pd_res.append([
get_q_text(sim.data[p_keys][-1], vocab),
number_dict[sp_text(sim.data[p_values][-1], vocab)],
np.sum(np.abs(sim.data[p_error]), axis=1)[-1],
np.max(spa.similarity(sim.data[p_recall], vocab))
])
print("Finished run %s of test %s" % (seed_val, t_n))
# Save as Pandas dataframe
base_name = "multpred2"
df = pd.DataFrame(pd_res, columns=pd_columns)
hdf = pd.HDFStore("results/%s_%s.h5" % (base_name, datetime.datetime.now().strftime("%I_%M_%S")))
df.to_hdf(hdf, base_name)