class LinearDecoder(Decoder):
"""
MLP Decoder for Hyperbolic/Euclidean node classification models.
"""
def __init__(self, c, args):
super(LinearDecoder, self).__init__(c)
self.manifold = getattr(manifolds, args.manifold)()
self.input_dim = args.dim
self.output_dim = args.n_classes
self.bias = args.bias
self.cls = Linear(self.input_dim, self.output_dim, args.dropout,
lambda x: x, self.bias)
self.decode_adj = False
def decode(self, x, adj):
h = self.manifold.proj_tan0(self.manifold.logmap0(x, c=self.c),
c=self.c)
return super(LinearDecoder, self).decode(h, adj)
def extra_repr(self):
return 'in_features={}, out_features={}, bias={}, c={}'.format(
self.input_dim, self.output_dim, self.bias, self.c)
def reset_parameters(self):
self.cls.reset_parameters()
print('GNN classification decoder reset finished')
def __init__(self, c, args):
super(LinearDecoder, self).__init__(c)
self.manifold = getattr(manifolds, args.manifold)()
self.input_dim = args.dim
self.output_dim = args.n_classes
self.bias = args.bias
self.cls = Linear(self.input_dim, self.output_dim, args.dropout,
lambda x: x, self.bias)
self.decode_adj = False
def __init__(self, c, args):
super(Shallow, self).__init__(c)
self.manifold = getattr(manifolds, args.manifold)()
self.use_feats = args.use_feats
weights = torch.Tensor(args.n_nodes, args.dim)
if not args.pretrained_embeddings:
weights = self.manifold.init_weights(weights, self.c)
trainable = True
else:
weights = torch.Tensor(np.load(args.pretrained_embeddings))
assert weights.shape[
0] == args.n_nodes, "The embeddings you passed seem to be for another dataset."
trainable = False
self.lt = manifolds.ManifoldParameter(weights, trainable,
self.manifold, self.c)
self.all_nodes = torch.LongTensor(list(range(args.n_nodes)))
layers = []
if args.pretrained_embeddings is not None and args.num_layers > 0:
# MLP layers after pre-trained embeddings
dims, acts = get_dim_act(args)
if self.use_feats:
dims[0] = args.feat_dim + weights.shape[1]
else:
dims[0] = weights.shape[1]
for i in range(len(dims) - 1):
in_dim, out_dim = dims[i], dims[i + 1]
act = acts[i]
layers.append(
Linear(in_dim, out_dim, args.dropout, act, args.bias))
self.layers = nn.Sequential(*layers)
self.encode_graph = False
def __init__(self, c, args):
super(LinearDecoder, self).__init__(c)
if args.manifold == 'MixedCurvature':
self.manifold = getattr(manifolds, args.manifold)(args.split_idx)
else:
self.manifold = getattr(manifolds, args.manifold)()
self.input_dim = args.dim
self.output_dim = args.n_classes
self.bias = args.bias
self.cls = Linear(self.input_dim, self.output_dim, args.dropout, lambda x: x, self.bias)
self.decode_adj = False
def __init__(self, c, args):
super(MLP, self).__init__(c)
assert args.num_layers > 0
dims, acts = get_dim_act(args)
layers = []
for i in range(len(dims) - 1):
in_dim, out_dim = dims[i], dims[i + 1]
act = acts[i]
layers.append(Linear(in_dim, out_dim, args.dropout, act, args.bias))
self.layers = nn.Sequential(*layers)
self.encode_graph = False
def __init__(self, c, args, task):
super(HGCAEDecoder, self).__init__(c)
self.manifold = getattr(manifolds, args.manifold)()
if task == 'nc':
self.input_dim = args.dim
self.output_dim = args.n_classes
self.bias = args.bias
self.classifier = Linear(self.input_dim, self.output_dim, args.dropout, lambda x: x, self.bias)
self.decode_adj = False
elif task == 'rec':
assert args.num_layers > 0
dims, acts, _ = hyp_layers.get_dim_act_curv(args)
dims = dims[::-1]
acts = acts[::-1][:-1] + [lambda x: x] # Last layer without act
self.curvatures = self.c[::-1]
encdec_share_curvature = False
if not encdec_share_curvature and args.num_layers == args.num_dec_layers: # do not share and enc-dec mirror-shape
num_c = len(self.curvatures)
self.curvatures = self.curvatures[:1]
if args.c_trainable == 1:
self.curvatures += [nn.Parameter(torch.Tensor([args.c]).to(args.device))] * (num_c - 1)
else:
self.curvatures += [torch.tensor([args.c])] * (num_c - 1)
if not args.cuda == -1:
self.curvatures = [curv.to(args.device) for curv in self.curvatures]
self.curvatures = self.curvatures[:-1] + [None]
hgc_layers = []
num_dec_layers = args.num_dec_layers
for i in range(num_dec_layers):
c_in, c_out = self.curvatures[i], self.curvatures[i + 1]
in_dim, out_dim = dims[i], dims[i + 1]
act = acts[i]
hgc_layers.append(
hyp_layers.HyperbolicGraphConvolution(
self.manifold, in_dim, out_dim, c_in, c_out, args.dropout, act, args.bias, args.use_att,
att_type=args.att_type, att_logit=args.att_logit, beta=args.beta, decode=True
)
)
self.decoder = nn.Sequential(*hgc_layers)
self.decode_adj = True
else:
raise RuntimeError('Unknown task')
def __init__(self, c, args, task):
super(HNNDecoder, self).__init__(c)
self.manifold = getattr(manifolds, args.manifold)()
if not args.cuda == -1:
c = torch.Tensor([c]).to(args.device)
if task == 'nc':
self.input_dim = args.dim
self.output_dim = args.n_classes
self.bias = args.bias
self.classifier = Linear(self.input_dim, self.output_dim, args.dropout, lambda x: x, self.bias)
self.decode_adj = False
elif task == 'rec':
assert args.num_layers > 0
dims, acts, _ = hyp_layers.get_dim_act_curv(args)
dims = dims[::-1]
acts = acts[::-1][:-1] + [lambda x: x] # Last layer without act
encdec_share_curvature = False
hnn_layers = []
num_dec_layers = args.num_dec_layers
for i in range(num_dec_layers):
in_dim, out_dim = dims[i], dims[i + 1]
act = acts[i]
c_in = c
c_out = None if (i == num_dec_layers - 1) else c
hnn_layers.append(
hyp_layers.HNNLayer(
self.manifold, in_dim, out_dim, c_in, c_out, args.dropout, act, args.bias
)
)
self.decoder = nn.Sequential(*hnn_layers)
self.decode_adj = False
else:
raise RuntimeError('Unknown task')
def __init__(self, c, args):
super(Shallow, self).__init__(c)
self.manifold = getattr(manifolds, args.manifold)()
self.use_feats = args.use_feats
self.pretrained_embeddings = args.pretrained_embeddings
self.n_nodes = args.n_nodes
self.weights = torch.Tensor(args.n_nodes, args.dim)
layers = []
if args.pretrained_embeddings is not None and args.num_layers > 0:
# MLP layers after pre-trained embeddings
dims, acts = get_dim_act(args)
if self.use_feats:
dims[0] = args.feat_dim + self.weights.shape[1]
else:
dims[0] = self.weights.shape[1]
for i in range(len(dims) - 1):
in_dim, out_dim = dims[i], dims[i + 1]
act = acts[i]
layers.append(
Linear(in_dim, out_dim, args.dropout, act, args.bias))
self.layers = nn.Sequential(*layers)
self.reset_parameteres()
self.encode_graph = False
