Multimodal demand forecasting aims at predicting product demand utilizing
visual, textual, and contextual information. This paper proposes a method for
multimodal product demand forecasting using convolutional, graph-based, and
transformer-based architectures. Traditional approaches to demand forecasting
rely on historical demand, product categories, and additional contextual
information such as seasonality and events. However, these approaches have
several shortcomings, such as the cold start problem making it difficult to
predict product demand until sufficient historical data is available for a
particular product, and their inability to properly deal with category
dynamics. By incorporating multimodal information, such as product images and
textual descriptions, our architecture aims to address the shortcomings of
traditional approaches and outperform them. The experiments conducted on a
large real-world dataset show that the proposed approach effectively predicts
demand for a wide range of products. The multimodal pipeline presented in this
work enhances the accuracy and reliability of the predictions, demonstrating
the potential of leveraging multimodal information in product demand
forecasting.

通过融合多模态信息，如产品图像和文字描述，提出一种使用卷积、基于图和 transformer 的架构进行多模态产品需求预测的方法，以解决传统方法的局限性并提高准确性和可靠性。实验证明，该方法对各类产品的需求预测有效。

多模态时间融合变压器是良好的产品需求预测器

Multimodal Temporal Fusion Transformers Are Good Product Demand  Forecasters

In online internet advertising, machine learning models are widely used to
compute the likelihood of a user engaging with product related advertisements.
However, the performance of traditional machine learning models is often
impacted due to variations in user and advertiser behavior. For example, search
engine traffic for florists usually tends to peak around Valentine's day,
Mother's day, etc. To overcome, this challenge, in this manuscript we propose
three models which are able to incorporate the effects arising due to
variations in product demand. The proposed models are a combination of product
demand features, specialized data sampling methodologies and ensemble
techniques. We demonstrate the performance of our proposed models on datasets
obtained from a real-world setting. Our results show that the proposed models
more accurately predict the outcome of users interactions with product related
advertisements while simultaneously being robust to fluctuations in user and
advertiser behaviors.

本文提出了三种结合产品需求特征、专业数据采样方法和集成技术的模型，可以更准确地预测产品广告的用户交互结果，并同时对用户和广告商行为的波动具有鲁棒性。