Module reclab.recommenders.autorec.autorec_lib.autorec

Classes

class AutoRec (num_users, num_items, seen_users, seen_items, hidden_neuron, dropout=0.05, random_seed=0)

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing to nest them in a tree structure. You can assign the submodules as regular attributes::

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self):
        super(Model, self).__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will have their parameters converted too when you call :meth:to, etc.

Initializes internal Module state, shared by both nn.Module and ScriptModule.

Expand source code

class AutoRec(torch.nn.Module):
    def __init__(self, num_users, num_items,
                 seen_users, seen_items,
                 hidden_neuron,
                 dropout=0.05, random_seed=0):
        super(AutoRec, self).__init__()
        self.num_users = num_users
        self.num_items = num_items
        self.seen_users = seen_users
        self.seen_items = seen_items

        self.hidden_neuron = hidden_neuron
        self.random_seed = random_seed
        self.dropout_p = dropout
        self.sigmoid = torch.nn.Sigmoid()

    def loss(self, pred, test, mask, lambda_value=1):
        mse = (((pred * mask) - test) ** 2).sum()
        reg_value_enc = torch.mul(lambda_value / 2, list(self.encoder.parameters())[0].norm(p='fro') ** 2)
        reg_value_dec = torch.mul(lambda_value / 2, list(self.decoder.parameters())[0].norm(p='fro') ** 2)
        return torch.add(mse, torch.add(reg_value_enc, reg_value_dec))

    def prepare_model(self):
        self.encoder = torch.nn.Linear(self.num_users, self.hidden_neuron, bias=True)
        self.dropout = torch.nn.Dropout(p=self.dropout_p)
        self.decoder = torch.nn.Linear(self.hidden_neuron, self.num_users, bias=True)

    def forward(self, x):
        x = self.encoder(x)
        x = self.sigmoid(x)
        x = self.dropout(x)
        x = self.decoder(x)
        return x

    def predict(self, user_item, test_data):
        users = [triple[0] for triple in user_item]
        items = [triple[1] for triple in user_item]

        user_item = zip(users, items)
        user_idx = set(users)
        item_idx = set(items)
        Estimated_R = self.forward(test_data)
        for item in range(test_data.shape[0]):
            for user in range(test_data.shape[1]):
                if user not in self.seen_users and item not in self.seen_items:
                    Estimated_R[item, user] = 3
        idx = [tuple(users), tuple(items)]
        Estimated_R = Estimated_R.clamp(1, 5)
        return Estimated_R.T[idx].cpu().detach().numpy()

Ancestors

• torch.nn.modules.module.Module

Methods

def forward(self, x)

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the :class:Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

Expand source code

def forward(self, x):
    x = self.encoder(x)
    x = self.sigmoid(x)
    x = self.dropout(x)
    x = self.decoder(x)
    return x

def loss(self, pred, test, mask, lambda_value=1)

Expand source code

def loss(self, pred, test, mask, lambda_value=1):
    mse = (((pred * mask) - test) ** 2).sum()
    reg_value_enc = torch.mul(lambda_value / 2, list(self.encoder.parameters())[0].norm(p='fro') ** 2)
    reg_value_dec = torch.mul(lambda_value / 2, list(self.decoder.parameters())[0].norm(p='fro') ** 2)
    return torch.add(mse, torch.add(reg_value_enc, reg_value_dec))

def predict(self, user_item, test_data)

Expand source code

def predict(self, user_item, test_data):
    users = [triple[0] for triple in user_item]
    items = [triple[1] for triple in user_item]

    user_item = zip(users, items)
    user_idx = set(users)
    item_idx = set(items)
    Estimated_R = self.forward(test_data)
    for item in range(test_data.shape[0]):
        for user in range(test_data.shape[1]):
            if user not in self.seen_users and item not in self.seen_items:
                Estimated_R[item, user] = 3
    idx = [tuple(users), tuple(items)]
    Estimated_R = Estimated_R.clamp(1, 5)
    return Estimated_R.T[idx].cpu().detach().numpy()

def prepare_model(self)

Expand source code

def prepare_model(self):
    self.encoder = torch.nn.Linear(self.num_users, self.hidden_neuron, bias=True)
    self.dropout = torch.nn.Dropout(p=self.dropout_p)
    self.decoder = torch.nn.Linear(self.hidden_neuron, self.num_users, bias=True)