Ing. Vojtěch Vančura, Ph.D.

Stať ve sborníku

10.1145/3705328.3748147

Katedra aplikované matematiky

Industry-scale recommender systems face a core challenge: representing entities with high cardinality, such as users or items, using dense embeddings that must be accessible during both training and inference. However, as embedding sizes grow, memory constraints make storage and access increasingly difficult. We describe a lightweight, learnable embedding compression technique that projects dense embeddings into a high-dimensional, sparsely activated space. Designed for retrieval tasks, our method reduces memory requirements while preserving retrieval performance, enabling scalable deployment under strict resource constraints. Our results demonstrate that leveraging sparsity is a promising approach for improving the efficiency of large-scale recommenders. We release our code at https://github.com/recombee/CompresSAE.

Stať ve sborníku

10.1145/3640457.3691707

Katedra aplikované matematiky

Recommender systems often use text-side information to improve their predictions, especially in cold-start or zero-shot recommendation scenarios, where traditional collaborative filtering approaches cannot be used. Many approaches to text-mining side information for recommender systems have been proposed over recent years, with sentence Transformers being the most prominent one. However, these models are trained to predict semantic similarity without utilizing interaction data with hidden patterns specific to recommender systems. In this paper, we propose beeFormer, a framework for training sentence Transformer models with interaction data. We demonstrate that our models trained with beeFormer can transfer knowledge between datasets while outperforming not only semantic similarity sentence Transformers but also traditional collaborative filtering methods. We also show that training on multiple datasets from different domains accumulates knowledge in a single model, unlocking the possibility of training universal, domain-agnostic sentence Transformer models to mine text representations for recommender systems. We release the source code, trained models, and additional details allowing replication of our experiments at https://github.com/recombee/beeformer.

Stať ve sborníku

10.1145/3565472.3595630

Katedra aplikované matematiky

The popularity of linear shallow autoencoders for collaborative filtering is growing in the research community, and internet industry providers of Recommender Systems are also taking notice. However, despite their simplicity and accuracy, these models often cannot be used in real-world industrial recommender systems due to their inability to scale to very large interaction matrices. Our research aims to address this issue by developing a scalable, explainable, and accurate shallow linear autoencoder method for collaborative filtering that meets the demands of real-world recommenders. In this paper, we present our industrial Ph.D. research project, which includes: (1) the development of a scalable method called ELSA and the adaptation of the method to a large real-world recommender and (2) the creation of a framework to visualize the recommender systems insights based on modeling the distribution of retrieval metrics in latent user space. We discuss the current status of our project, the key steps to finish the project, and the possible future extensions after the dissertation.

Stať ve sborníku

10.1109/CSIT56902.2022.10000782

Katedra softwarového inženýrství
Katedra aplikované matematiky
Fakulta informačních technologií

Offline optimization of recommender systems is a difficult task. Popular optimization criteria such as RMSE, Recall, and NDCG do not correlate much with online performance, especially when the recommendation algorithm is largely different from the one used to generate the offline data. An exciting direction of research to mitigate this problem is to use more robust optimization criteria. Serendipity is reported to be a promising proxy. However, more variants exist, and it is unclear whether they can be used as a single criterion to optimize. This paper analyzes how serendipity relates to other optimization criteria for three different recommendation algorithms. Based on our findings, we propose to modify the way serendipity is computed. We conduct experiments using three collaborative filtering algorithms: K-Nearest Neighbors, Matrix Factorization, and Embarrassingly Shallow Autoencoder (EASE). We also employ and evaluate the ensemble learning approach and analyze the importance of the individual components of serendipity.

Stať ve sborníku

10.1145/3523227.3551469

Katedra aplikované matematiky

Making recommender systems more transparent and auditable is crucial for the future adoption of these systems. Available tools typically present mostly errors of models aggregated over all test users, which is often insufficient to uncover hidden biases and problems. Moreover, the emphasis is primarily on the accuracy of recommendations but less on other important metrics, such as the diversity of recommended items, the extent of catalog coverage, or the opportunity to discover novel items at bestsellers’ expense. In this work, we propose RepSys, a framework for evaluating recommender systems. Our work offers a set of highly interactive approaches for investigating various scenario recommendations, analyzing a dataset, and evaluating distributions of various metrics that combine visualization techniques with existing offline evaluation methods. RepSys framework is available under an open-source license to other researchers.

Stať ve sborníku

10.1145/3523227.3551482

Katedra aplikované matematiky

Recently, the RS research community has witnessed a surge in popularity for shallow autoencoder-based CF methods. Due to its straightforward implementation and high accuracy on item retrieval metrics, EASE is potentially the most prominent of these models. Despite its accuracy and simplicity, EASE cannot be employed in some real-world recommender system applications due to its inability to scale to huge interaction matrices. In this paper, we proposed ELSA, a scalable shallow autoencoder method for implicit feedback recommenders. ELSA is a scalable autoencoder in which the hidden layer is factorizable into a low-rank plus sparse structure, thereby drastically lowering memory consumption and computation time. We conducted a comprehensive offline experimental section that combined synthetic and several real-world datasets. We also validated our strategy in an online setting by comparing ELSA to baselines in a live recommender system using an A/B test. Experiments demonstrate that ELSA is scalable and has competitive performance. Finally, we demonstrate the explainability of ELSA by illustrating the recovered latent space.

Stať ve sborníku

10.1007/978-3-030-86383-8_11

Katedra aplikované matematiky

The recently introduced Embarrasingelly Shallow Autoencoder (EASE) algorithm presents a simple and elegant way to solve the top-N recommendation task. In this paper, we introduce Neural EASE to further improve the performance of this algorithm by incorporating techniques for training modern neural networks. Also, there is a growing interest in the recsys community to utilize variational autoencoders (VAE) for this task. We introduce Focal Loss Variational AutoEncoder (FLVAE), benefiting from multiple non-linear layers without an information bottleneck while not overfitting towards the identity. We show how to learn FLVAE in parallel with Neural EASE and achieve state-of-the-art performance on the MovieLens 20M dataset and competitive results on the Netflix Prize dataset.

Stať ve sborníku

10.1145/3460231.3473896

Katedra aplikované matematiky

Next basket prediction from historical purchases is quite a complex task, even for e-commerce datasets with a low number of items that are being purchased repeatedly. Neural approaches are not much better in predicting next purchases than simple heuristics. This paper focuses on the challenge of how to encode baskets into efficient neural embedding with low reconstruction error while maintaining the similarity of baskets in the latent space. In our representation, replacing a product with a similar product or increasing quantity will not change the embedding of the basket much. We believe that good basket representation is critical for subsequent prediction. Our analysis shows that state-of-the-art next basket prediction approaches have limitations in their representation of baskets. We would like to focus on this aspect in our future research.