def test_add_new_element(self):
om = ag.ObjectsMemory()
t1 = torch.arange(10, 20)[None, ...]
om.add_new_element(t1, 1)
npt.assert_equal(utils.t2a(om.M), np.arange(10, 20).reshape((1, 10)))
npt.assert_equal(om.seq_ids, np.arange(1, 2))
g_set = set([1])
self.assertEqual(set(om.G.nodes), g_set)
om = ag.ObjectsMemory()
t1 = torch.arange(10, 20)[None, ...]
t2 = torch.arange(20, 30)[None, ...]
t3 = torch.arange(30, 40)[None, ...]
om.add_new_element(t1, 1)
om.add_new_element(t2, 2)
om.add_new_element(t3, 3)
npt.assert_equal(utils.t2a(om.M), np.arange(10, 40).reshape((3, 10)))
npt.assert_equal(om.seq_ids, np.arange(1, 4))
g_set = set([1, 2, 3])
self.assertEqual(set(om.G.nodes), g_set)
def get_data(session_ds, inc_eval_ds, ms_band, db_eps):
session_data = list(session_ds)
inc_eval_data = list(inc_eval_ds)
session_emb = np.squeeze([utils.t2a(d[0][0]) for d in session_data])
session_lab = np.squeeze([d[1] for d in session_data])
inc_eval_emb = np.squeeze([utils.t2a(d[0][0]) for d in inc_eval_data])
inc_eval_lab = np.squeeze([d[1] for d in inc_eval_data])
X = np.concatenate((session_emb, inc_eval_emb))
y = np.concatenate((session_lab, inc_eval_lab))
meanshifts = [cl.MeanShift(bandwidth=b).fit_predict(X) for b in ms_band]
optics = cl.OPTICS(min_samples=1).fit_predict(X)
dbscans = [cl.DBSCAN(eps=e, min_samples=1).fit_predict(X) for e in db_eps]
res = np.array(meanshifts + dbscans + [optics])
inc_pred = res[:, session_lab.size:]
aris = [adjusted_rand_score(p, inc_eval_lab) for p in inc_pred]
amis = [
adjusted_mutual_info_score(p, inc_eval_lab, average_method='max')
for p in inc_pred
]
return np.array(aris), np.array(amis), inc_pred, inc_eval_lab
def test_get_something(self):
om = ag.ObjectsMemory()
t1 = torch.arange(10, 20)
t2 = torch.arange(20, 30)
t3 = torch.arange(30, 40)
om.add_new_element(t1, 1)
om.add_new_element(t2, 2)
om.add_new_element(t3, 3)
npt.assert_equal(om.get_sid(0), 1)
npt.assert_equal(utils.t2a(om.get_embed(0)), utils.t2a(t1))
def test_globalmean(self):
bogusdata = [torch.ones(5, 3), torch.zeros(6, 3)]
gm = models.GlobalMean()
fded = gm.forward(bogusdata)
fded = np.array([utils.t2a(f.squeeze(dim=0)) for f in fded])
npt.assert_equal(fded, np.array([np.ones(3), np.zeros(3)]))
fded = gm.forward(bogusdata[:1])
fded = utils.t2a(fded[0].squeeze(dim=0))
npt.assert_equal(fded, np.ones(3))
def test_recursivereduction(self):
bogusdata = [torch.ones(50, 3), torch.zeros(60, 3)]
rr = models.RecursiveReduction(3)
fded = rr.forward(bogusdata)
fded_a = np.concatenate([utils.t2a(f) for f in fded])
self.assertEqual(fded_a.shape, (2, 3))
def test_get_knn(self):
om = ag.ObjectsMemory()
t1 = torch.arange(10, 20).float()
t2 = torch.arange(20, 30).float()
t3 = torch.arange(40, 50).float()
om.add_new_element(t1, 1)
om.add_new_element(t2, 2)
om.add_new_element(t3, 3)
npt.assert_equal(utils.t2a(om.get_knn(t1, k=1)[1]), 0)
npt.assert_equal(utils.t2a(om.get_knn(t1, k=2)[1]), np.array([[0, 1]]))
t12 = torch.stack([t1, t2])
npt.assert_equal(utils.t2a(om.get_knn(t12, k=1)[1]),
np.array([[0], [1]]))
npt.assert_equal(utils.t2a(om.get_knn(t12, k=2)[1]),
np.array([[0, 1], [1, 0]]))
def test_add_neighbors(self):
om = ag.ObjectsMemory()
t1 = torch.arange(10, 20)[None, ...]
t2 = torch.arange(20, 30)[None, ...]
t3 = torch.arange(30, 40)[None, ...]
om.add_new_element(t1, 1)
om.add_new_element(t2, 2)
om.add_new_element(t3, 3)
om.add_neighbors(3, [0])
npt.assert_equal(utils.t2a(om.M), np.arange(10, 40).reshape((3, 10)))
npt.assert_equal(om.seq_ids, np.arange(1, 4))
g_set = set([1, 2, 3])
self.assertEqual(set(om.G.nodes), g_set)
e_set = set([(1, 3)])
self.assertEqual(set([tuple(sorted(e)) for e in om.G.edges]), e_set)
def t_assert_equal(a, b):
npt.assert_equal(utils.t2a(a), utils.t2a(b))
