A Meta-Learning Perspective on Cold-Start Recommendations for Items

By Manasi Vartak, Arvind Thiagarajan, Conrado Silva Miranda, Jeshua Bratman, Hugo Larochelle. 2017

232 citations recorded by the source corpus.

Subfields: Meta & Continual Learning, Recommender Systems

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Meta & Continual Learning

In Recommender Systems

Citations in the corpus

Fast and Flexible Multi-Task Classification Using Conditional Neural Adaptive Processes (2019, background): “Third, as permutation invariant architectures can be incrementally updated [20], continual learning is supported (as discussed in Section 5).”
Meta-Learning: A Survey (2018, neutral): “Finally, Vartak et al. [177] address the cold-start problem in matrix factorization.”
Federated Meta-Learning with Fast Convergence and Efficient Communication (2018, methodology): “ommender systems, and propose a general federated meta-learning framework that can cope with a wide range of meta-learning algorithms as well as recommendation models. Secondly, the architectures in (Vartak et al., 2017) share user information at the model level with part of the network encoding met”
Generalizing from a Few Examples (2019, background): “s to be considered in other fields. In natural language processing, examples are translation [56] and language modeling [127]. In recommendation, cold-start item recommendation has been considered in [126]. In medical applications, few-shot drug discovery is attempted in [3]. One-shot architecture s”
Few-shot learning with adaptively initialized task optimizer: a practical meta-learning approach (2019, background): “2018a), recommendation system (Vartak et al. 2017), and hyper-parameter tuning (Franceschi et al.”
How to Train Your MAML to Excel in Few-Shot Classification (2021, methodology): “Meta-learning has been applied in various fields, including imbalance learning [74, 106], data compression [103], architecture search [26], recommendation systems [96], data augmentation [69], teaching [27], and hyper-parameter tuning [33, 66].”
Contextual Explanation Networks (2017, background): “Meta-training a deep network that generates parameters for another network has been successfully used for zero-shot [19, 20] and few-shot [21, 22] learning, cold-start recommendations [23], and a few other scenarios [24, 25, 26], but is not suitable for interpretability purposes.”
Federated Meta-Learning for Recommendation (2018, methodology): “Vartak et al. (Vartak et al., 2017) studied cold-start recommendation for items from a meta-learning perspective.”
Revisiting Meta-Learning as Supervised Learning (2020, background): “, 2017), recommendation systems (Vartak et al., 2017), data augmentation (Ratner et al.”
Deep Learning Based Recommender System (2017, unknown): “Others Cold-start [154, 156, 170, 171] Multitask [5, 73, 87, 174, 187] Explainability [87, 126]”
Adversarially Robust Few-Shot Learning: A Meta-Learning Approach (2019, background): “When data is scarce or new classes arise frequently, neural networks must adapt quickly [7, 14, 24, 30].”
Meta-learning on Heterogeneous Information Networks for Cold-start Recommendation (2020, background): “On another line, at the model level, the recent episodic metalearning paradigm [9] has offered insights into modeling new users or items with scarce interaction data [27].”
Latent Cross: Making Use of Context in Recurrent Recommender Systems (2018, background): “More complex model structures like attention models [3], memory networks [38], and meta-learning [42] also rely on second-order relations and are increasingly popular.”
Contrastive Learning for Cold-Start Recommendation (2021, background): “Technically, they augment the training data by randomly corrupting warm items’ collaborative embeddings [8, 38, 43] or manually crafting new items [25, 31, 34, 42], so as to enhance the generalization ability to unseen items.”
Robust Few-Shot Learning with Adversarially Queried Meta-Learners (2019, background): “When data is scarce or new classes arise frequently, neural networks must adapt quickly (Duan et al., 2017; Kaiser et al., 2017; Pfister et al., 2014; Vartak et al., 2017).”
MeLU: Meta-Learned User Preference Estimator for Cold-Start Recommendation (2019, background): “Previous researches [8, 20, 22, 27] on metric-based and memory-based meta-learning are concentrated on classification problems; the work in [25] took a concept of the metric-based meta-learning algorithm in the recommender system to predict whether a user consumes an item or not.”