def test_map(self):
nn = 50
_x = np.random.normal(size=(nn, 3))
_y = .4 * _x[:,
0] + 0.2 * _x[:,
1] - 0.3 * _x[:,
2] + 1.5 + np.random.normal(
size=(nn), scale=.1)
_x = tf.convert_to_tensor(_x)
_y = tf.convert_to_tensor(_y)
model = tf.keras.Sequential([tf.keras.layers.Dense(1)])
def _make_prior(posterior):
n = len(posterior.event_shape)
return tfd.Independent(tfd.Normal(
tf.zeros(posterior.event_shape, dtype=posterior.dtype),
tf.constant(2., dtype=posterior.dtype)),
reinterpreted_batch_ndims=n)
bayesian_model = svi.SVI(model,
kl_scale=1.0,
prior_fn=_make_prior,
posterior_fn=svi.make_deterministic_posterior)
bayesian_model.compile(
loss=tfk.losses.MeanSquaredError(),
optimizer=tf.keras.optimizers.Adam(learning_rate=0.01),
run_eagerly=False)
bayesian_model.fit(_x, _y, epochs=2, batch_size=nn, verbose=1)
return
def test_mc_kl(self):
nn = 50
_x = np.random.normal(size=(nn, 3))
_y = .4 * _x[:,
0] + 0.2 * _x[:,
1] - 0.3 * _x[:,
2] + 1.5 + np.random.normal(
size=(nn), scale=.1)
_x = tf.convert_to_tensor(_x)
_y = tf.convert_to_tensor(_y)
model = tf.keras.Sequential(
[tf.keras.Input(shape=(3)),
tf.keras.layers.Dense(1)])
bayesian_model = svi.SVI(model,
kl_scale=1.0,
kl_fn=svi.mc_kl,
prior_fn=svi.make_spike_and_slab_prior,
posterior_fn=svi.make_mvn_posterior)
bayesian_model.compile(
loss=tfk.losses.MeanSquaredError(),
optimizer=tf.keras.optimizers.Adam(learning_rate=0.01))
bayesian_model.fit(_x, _y, epochs=2, batch_size=nn, verbose=1)
def test_rnn(self):
x = np.random.normal(size=(200, 10, 1))
y = np.random.normal(size=(200, 1))
model = tf.keras.Sequential([
tf.keras.Input(shape=(10, 1)),
# tf.keras.layers.GRU(4),
tf.keras.layers.RNN(tfk.layers.GRUCell(4)),
tf.keras.layers.Dense(1)
])
bayesian_model = svi.SVI(model, kl_scale=1.0)
bayesian_model.compile(
loss=tfk.losses.MeanSquaredError(),
optimizer=tf.keras.optimizers.Adam(learning_rate=0.01))
bayesian_model.fit(x, y, epochs=2, batch_size=64, verbose=1)
def test_built(self):
nn = 50
_x = np.random.normal(size=(nn, 3)).astype(np.float32)
_y = .4 * _x[:,
0] + 0.2 * _x[:,
1] - 0.3 * _x[:, 2] + 1.5 + np.random.normal(
size=(nn), scale=.1).astype(np.float32)
_x = tf.convert_to_tensor(_x)
_y = tf.convert_to_tensor(_y)
model = tf.keras.Sequential([tf.keras.layers.Dense(1)])
model(_x)
bayesian_model = svi.SVI(model, kl_scale=1.0)
bayesian_model.compile(
loss=tfk.losses.MeanSquaredError(),
optimizer=tf.keras.optimizers.Adam(learning_rate=0.01))
bayesian_model.fit(_x, _y, epochs=2, batch_size=nn, verbose=1)
return
def test_test_step(self):
nn = 50
_x = np.random.normal(size=(nn, 3))
_y = .4 * _x[:,
0] + 0.2 * _x[:,
1] - 0.3 * _x[:,
2] + 1.5 + np.random.normal(
size=(nn), scale=.1)
_x = tf.convert_to_tensor(_x, dtype=tf.keras.backend.floatx())
_y = tf.convert_to_tensor(_y, dtype=tf.keras.backend.floatx())
model = tf.keras.Sequential([
tf.keras.layers.Dense(1),
tfp.layers.DistributionLambda(
lambda t: tfd.Independent(tfd.Normal(loc=t, scale=t**2), 1))
])
bayesian_model = svi.SVI(model,
kl_scale=1.0,
kl_fn=svi.mc_kl,
prior_fn=svi.make_spike_and_slab_prior,
posterior_fn=svi.make_mvn_posterior)
bayesian_model.compile(
loss=lambda y, rv_y: -rv_y.log_prob(y),
optimizer=tf.keras.optimizers.Adam(learning_rate=0.01),
metrics=['mse'])
h = bayesian_model.fit(_x,
_y,
epochs=2,
batch_size=nn,
validation_data=(_x, _y),
verbose=1)
self.assertTrue('val_loss' in h.history)
self.assertTrue('val_mse' in h.history)
pass