FEDIN: Frequency-Enhanced Deep Interest Network for Click-Through Rate Prediction

TL;DR

FEDIN is a novel recommendation model that leverages frequency-domain analysis to better capture periodic user interests and improve click-through rate prediction accuracy.

Contribution

It introduces a frequency-domain branch with target-aware spectrum filtering, effectively isolating periodic interest signals in sequential recommendation.

Findings

FEDIN outperforms state-of-the-art baselines on three datasets.

The frequency-domain approach enhances robustness against noise.

Spectral entropy distributions differ significantly between positive and negative targets.

Abstract

Sequential recommendation models often struggle to capture latent periodic patterns in user interests, primarily due to the noise inherent in time-domain behavioral data. While frequency-domain analysis offers a global perspective to address this, existing approaches typically treat user sequences in isolation, overlooking the crucial context of the target item. In this work, we present a novel empirical observation: user attention scores exhibit distinct spectral entropy distributions when conditioned on positive versus negative target items. Specifically, true user interests manifest as highly concentrated spectral patterns with lower entropy in the frequency domain, whereas irrelevant behaviors appear as high-entropy noise. Leveraging this insight, we propose the Frequency-Enhanced Deep Interest Network (FEDIN). FEDIN introduces a frequency-domain branch that utilizes a target-aware spectrum filtering mechanism to isolate these periodic interest signals. Extensive experiments on three public datasets demonstrate that FEDIN consistently outperforms state-of-the-art sequential recommendation baselines, demonstrating superior robustness against noise. We have released our code at: https://github.com/otokoneko/FEDIN.