Skip to content

Commit

Permalink
Add files via upload
Browse files Browse the repository at this point in the history
  • Loading branch information
@benedekrozemberczki
benedekrozemberczki authored Dec 29, 2018
1 parent d8b4d3f commit 502fcff
Show file tree
Hide file tree
Showing 3 changed files with 80 additions and 80 deletions.
14 changes: 7 additions & 7 deletions src/calculation_helper.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -30,7 +30,7 @@ def overlap(g, node_1, node_2):

def unit(g, node_1, node_2):
"""
Function to calculate the 'unit' weight.
Function to calculate the "unit" weight.
:param g: NX graph.
:param node_1: Node 1. of a pair.
:param node_2: Node 2. of a pair.
Expand Down Expand Up @@ -181,8 +181,8 @@ def __init__(self, nx_G, is_directed, p, q):
self.nodes = nx.nodes(self.G)
print("Edge weighting.\n")
for edge in tqdm(self.G.edges()):
self.G[edge[0]][edge[1]]['weight'] = 1.0
self.G[edge[1]][edge[0]]['weight'] = 1.0
self.G[edge[0]][edge[1]]["weight"] = 1.0
self.G[edge[1]][edge[0]]["weight"] = 1.0
self.is_directed = is_directed
self.p = p
self.q = q
Expand Down Expand Up @@ -249,11 +249,11 @@ def get_alias_edge(self, src, dst):
unnormalized_probs = []
for dst_nbr in sorted(G.neighbors(dst)):
if dst_nbr == src:
unnormalized_probs.append(G[dst][dst_nbr]['weight']/p)
unnormalized_probs.append(G[dst][dst_nbr]["weight"]/p)
elif G.has_edge(dst_nbr, src):
unnormalized_probs.append(G[dst][dst_nbr]['weight'])
unnormalized_probs.append(G[dst][dst_nbr]["weight"])
else:
unnormalized_probs.append(G[dst][dst_nbr]['weight']/q)
unnormalized_probs.append(G[dst][dst_nbr]["weight"]/q)
norm_const = sum(unnormalized_probs)
normalized_probs = [float(u_prob)/norm_const for u_prob in unnormalized_probs]

Expand All @@ -270,7 +270,7 @@ def preprocess_transition_probs(self):
print("")
print("Preprocesing.\n")
for node in tqdm(G.nodes()):
unnormalized_probs = [G[node][nbr]['weight'] for nbr in sorted(G.neighbors(node))]
unnormalized_probs = [G[node][nbr]["weight"] for nbr in sorted(G.neighbors(node))]
norm_const = sum(unnormalized_probs)
normalized_probs = [float(u_prob)/norm_const for u_prob in unnormalized_probs]
alias_nodes[node] = alias_setup(normalized_probs)
Expand Down
138 changes: 69 additions & 69 deletions src/parser.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -13,137 +13,137 @@ def parameter_parser():
# Input and output file parameters.
#------------------------------------------------------------------

parser.add_argument('--input',
nargs = '?',
default = './data/politician_edges.csv',
help = 'Input graph path.')

parser.add_argument('--embedding-output',
nargs = '?',
default = './output/embeddings/politician_embedding.csv',
help = 'Embeddings path.')

parser.add_argument('--cluster-mean-output',
nargs = '?',
default = './output/cluster_means/politician_means.csv',
help = 'Cluster means path.')

parser.add_argument('--log-output',
nargs = '?',
default = './output/logs/politician.json',
help = 'Log path.')

parser.add_argument('--assignment-output',
nargs = '?',
default = './output/assignments/politician.json',
help = 'Log path.')

parser.add_argument('--dump-matrices',
parser.add_argument("--input",
nargs = "?",
default = "./data/politician_edges.csv",
help = "Input graph path.")

parser.add_argument("--embedding-output",
nargs = "?",
default = "./output/embeddings/politician_embedding.csv",
help = "Embeddings path.")

parser.add_argument("--cluster-mean-output",
nargs = "?",
default = "./output/cluster_means/politician_means.csv",
help = "Cluster means path.")

parser.add_argument("--log-output",
nargs = "?",
default = "./output/logs/politician.json",
help = "Log path.")

parser.add_argument("--assignment-output",
nargs = "?",
default = "./output/assignments/politician.json",
help = "Log path.")

parser.add_argument("--dump-matrices",
type = bool,
default = True,
help = 'Save the embeddings to disk or not. Default is not.')
help = "Save the embeddings to disk or not. Default is not.")

parser.add_argument('--model',
nargs = '?',
default = 'GEMSECWithRegularization',
help = 'The model type.')
parser.add_argument("--model",
nargs = "?",
default = "GEMSECWithRegularization",
help = "The model type.")

#------------------------------------------------------------------
# RandomWalk parameters.
#------------------------------------------------------------------

parser.add_argument('--P',
parser.add_argument("--P",
type = float,
default = 1,
help = 'Return hyperparameter. Default is 1.')
help = "Return hyperparameter. Default is 1.")

parser.add_argument('--Q',
parser.add_argument("--Q",
type = float,
default = 1,
help = 'In-out hyperparameter. Default is 1.')
help = "In-out hyperparameter. Default is 1.")

parser.add_argument('--walker',
nargs = '?',
default = 'first',
help = 'Random walker order. Default is first.')
parser.add_argument("--walker",
nargs = "?",
default = "first",
help = "Random walker order. Default is first.")


#------------------------------------------------------------------
# Skipgram parameters.
#------------------------------------------------------------------

parser.add_argument('--dimensions',
parser.add_argument("--dimensions",
type = int,
default = 16,
help = 'Number of dimensions. Default is 16.')
help = "Number of dimensions. Default is 16.")

parser.add_argument('--random-walk-length',
parser.add_argument("--random-walk-length",
type = int,
default = 80,
help = 'Length of random walk per source. Default is 80.')
help = "Length of random walk per source. Default is 80.")

parser.add_argument('--num-of-walks',
parser.add_argument("--num-of-walks",
type = int,
default = 5,
help = 'Number of random walks per source. Default is 5.')
help = "Number of random walks per source. Default is 5.")

parser.add_argument('--window-size',
parser.add_argument("--window-size",
type = int,
default = 5,
help = 'Window size for proximity statistic extraction. Default is 5.')
help = "Window size for proximity statistic extraction. Default is 5.")

parser.add_argument('--distortion',
parser.add_argument("--distortion",
type = float,
default = 0.75,
help = 'Downsampling distortion. Default is 0.75.')
help = "Downsampling distortion. Default is 0.75.")

parser.add_argument('--negative-sample-number',
parser.add_argument("--negative-sample-number",
type = int,
default = 10,
help = 'Number of negative samples to draw. Default is 10.')
help = "Number of negative samples to draw. Default is 10.")

#------------------------------------------------------------------
# Optimization parameters.
#------------------------------------------------------------------

parser.add_argument('--initial-learning-rate',
parser.add_argument("--initial-learning-rate",
type = float,
default = 0.01,
help = 'Initial learning rate. Default is 0.01.')
help = "Initial learning rate. Default is 0.01.")

parser.add_argument('--minimal-learning-rate',
parser.add_argument("--minimal-learning-rate",
type = float,
default = 0.001,
help = 'Minimal learning rate. Default is 0.001.')
help = "Minimal learning rate. Default is 0.001.")

parser.add_argument('--annealing-factor',
parser.add_argument("--annealing-factor",
type = float,
default = 1,
help = 'Annealing factor. Default is 1.0.')
help = "Annealing factor. Default is 1.0.")

parser.add_argument('--initial-gamma',
parser.add_argument("--initial-gamma",
type = float,
default = 0.1,
help = 'Initial clustering weight. Default is 0.1.')
help = "Initial clustering weight. Default is 0.1.")

parser.add_argument('--lambd',
parser.add_argument("--lambd",
type = float,
default = 2.0**-4,
help = 'Smoothness regularization penalty. Default is 0.0625.')
help = "Smoothness regularization penalty. Default is 0.0625.")

parser.add_argument('--cluster-number',
parser.add_argument("--cluster-number",
type = int,
default = 20,
help = 'Number of clusters. Default is 20.')
help = "Number of clusters. Default is 20.")

parser.add_argument('--overlap-weighting',
nargs = '?',
default = 'normalized_overlap',
help = 'Weight construction technique for regularization.')
parser.add_argument("--overlap-weighting",
nargs = "?",
default = "normalized_overlap",
help = "Weight construction technique for regularization.")

parser.add_argument('--regularization-noise',
parser.add_argument("--regularization-noise",
type = float,
default = 10**-8,
help = 'Uniform noise max and min on the feature vector distance.')
help = "Uniform noise max and min on the feature vector distance.")

return parser.parse_args()
8 changes: 4 additions & 4 deletions src/print_and_read.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -26,7 +26,7 @@ def json_dumper(data, path):
"""
Function to dump the logs and assignments.
"""
with open(path, 'w') as outfile:
with open(path, "w") as outfile:
json.dump(data, outfile)

def initiate_dump_gemsec(log, assignments, args, final_embeddings, c_means):
Expand Down Expand Up @@ -56,15 +56,15 @@ def tab_printer(log):
Function to print the logs in a nice tabular format.
"""
t = Texttable()
t.add_rows([['Epoch', log["losses"][-1][0]]])
t.add_rows([["Epoch", log["losses"][-1][0]]])
print(t.draw())

t = Texttable()
t.add_rows([['Loss', round(log["losses"][-1][1],3)]])
t.add_rows([["Loss", round(log["losses"][-1][1],3)]])
print(t.draw())

t = Texttable()
t.add_rows([['Modularity', round(log["cluster_quality"][-1][1],3)]])
t.add_rows([["Modularity", round(log["cluster_quality"][-1][1],3)]])
print(t.draw())

def epoch_printer(repetition):
Expand Down

0 comments on commit 502fcff

Please sign in to comment.