RaCT

Reference:

Sam Lobel et al. “RaCT: Towards Amortized Ranking-Critical Training for Collaborative Filtering.” in ICLR 2020.

class recbole.model.general_recommender.ract.RaCT(config, dataset)

Bases: GeneralRecommender, AutoEncoderMixin

RaCT is a collaborative filtering model which uses methods based on actor-critic reinforcement learning for training.

We implement the RaCT model with only user dataloader.

calculate_ac_loss(interaction)

calculate_actor_loss(interaction)

calculate_critic_loss(interaction)

calculate_loss(interaction)

Calculate the training loss for a batch data.

Parameters:

interaction (Interaction) – Interaction class of the batch.

Returns:

Training loss, shape: []

Return type:

torch.Tensor

calculate_ndcg(predict_matrix, true_matrix, input_matrix, k)

construct_critic_input(actor_loss)

construct_critic_layers(layer_dims)

critic_forward(actor_loss)

forward(rating_matrix)

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 Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

full_sort_predict(interaction)

full sort prediction function. Given users, calculate the scores between users and all candidate items.

Parameters:

interaction (Interaction) – Interaction class of the batch.

Returns:

Predicted scores for given users and all candidate items, shape: [n_batch_users * n_candidate_items]

Return type:

torch.Tensor

input_type = 2

mlp_layers(layer_dims)

predict(interaction)

Predict the scores between users and items.

Parameters:

interaction (Interaction) – Interaction class of the batch.

Returns:

Predicted scores for given users and items, shape: [batch_size]

Return type:

torch.Tensor

reparameterize(mu, logvar)

training: bool