• Corpus ID: 244728320

Lightweight representation learning for efficient and scalable recommendation

@inproceedings{Koch2021LightweightRL,
  title={Lightweight representation learning for efficient and scalable recommendation},
  author={Olivier Koch and Amine Benhalloum and Guillaume Genthial and Denis Kuzin and Dmitry Parfenchik},
  year={2021}
}
Over the past decades, recommendation has become a critical component of many online services such as media streaming and ecommerce. Billion-scale recommendation engines are becoming common place. Meanwhile, new state-of-the-art methods such as variational auto-encoders have appeared, raising the bar in performance. We propose to go a step further with a simple and efficient model (LED, for Lightweight Encoder-Decoder). By combining pretraining, sampled losses and amortized inference, LED… 

References

SHOWING 1-10 OF 48 REFERENCES
LightRec: A Memory and Search-Efficient Recommender System
TLDR
This work proposes LightRec, a lightweight recommender system which enjoys fast online inference and economic memory consumption, and devise an end-to-end learning workflow, where challenges on the inherent differentiability and diversity are conquered by the proposed techniques.
Graph Convolutional Neural Networks for Web-Scale Recommender Systems
TLDR
A novel method based on highly efficient random walks to structure the convolutions and a novel training strategy that relies on harder-and-harder training examples to improve robustness and convergence of the model are developed.
Neural Input Search for Large Scale Recommendation Models
TLDR
This work proposes a novel type of embedding, namely Multi-size Embedding (ME), which allows the embedding dimension to vary for different values of the feature, and argues that the traditional Single-size embedding (SE) suffers from inefficient usage of model capacity and training data.
Deep content-based music recommendation
TLDR
This paper proposes to use a latent factor model for recommendation, and predict the latent factors from music audio when they cannot be obtained from usage data, and shows that recent advances in deep learning translate very well to the music recommendation setting, with deep convolutional neural networks significantly outperforming the traditional approach.
Temporal-Contextual Recommendation in Real-Time
TLDR
This work presents a black-box recommender system that can adapt to a diverse set of scenarios without the need for manual tuning, and introduces a compact model, which is called hierarchical recurrent network with meta data (HRNN-meta) to address the real-time and diverse metadata needs.
Deep Interest Network for Click-Through Rate Prediction
TLDR
A novel model: Deep Interest Network (DIN) is proposed which tackles this challenge by designing a local activation unit to adaptively learn the representation of user interests from historical behaviors with respect to a certain ad.
IntentGC: A Scalable Graph Convolution Framework Fusing Heterogeneous Information for Recommendation
TLDR
This work collects abundant relationships from common user behaviors and item information, and proposes a novel framework named IntentGC to leverage both explicit preferences and heterogeneous relationships by graph convolutional networks to solve the sparsity of user-item interactions on websites.
RecVAE: A New Variational Autoencoder for Top-N Recommendations with Implicit Feedback
TLDR
The Recommender VAE (RecVAE) model is proposed that significantly outperforms previously proposed autoencoder-based models, including Mult-VAE and RaCT, across classical collaborative filtering datasets, and is presented with a detailed ablation study to assess new developments.
Use of Deep Learning in Modern Recommendation System: A Summary of Recent Works
TLDR
The recent advances made in the field of recommendation using various variants of deep learning technology are covered and whether deep learning has shown any significant improvement over the conventional systems for recommendation is discussed.
Variational Autoencoders for Collaborative Filtering
TLDR
A generative model with multinomial likelihood and use Bayesian inference for parameter estimation is introduced and the pros and cons of employing a principledBayesian inference approach are identified and characterize settings where it provides the most significant improvements.
...
...