Explainable Multi-Agent Recommendation System for Energy-Efﬁcient Decision Support in Smart Homes

2025-05-06 0 0 597.19KB 9 页 10玖币

侵权投诉

Explainable Multi-Agent Recommendation System

for Energy-Efﬁcient Decision Support in Smart

Homes

Alona Zharova∗

Humboldt-Universität zu Berlin

Berlin, Germany

alona.zharova@hu-berlin.de

Annika Boer

Humboldt-Universität zu Berlin

Berlin, Germany

boeranni@hu-berlin.de

Julia Knoblauch

Humboldt-Universität zu Berlin

Berlin, Germany

julia.knoblauch@hu-berlin.de

Kai Ingo Schewina

Humboldt-Universität zu Berlin

Berlin, Germany

schewink@hu-berlin.de

Jana Vihs

Humboldt-Universität zu Berlin

Berlin, Germany

vihsjana@hu-berlin.de

Abstract

Understandable and persuasive recommendations support the electricity consumers’

behavioral change to tackle the energy efﬁciency problem. This paper proposes an

explainable multi-agent recommendation system for load shifting for the household

appliances. First, we provide agents with enhanced predictive capacity by including

weather data, applying state-of-the-art models and tuning the hyperparameters. Sec-

ond, we suggest an Explainability Agent providing transparent recommendations.

Third, we discuss the potential impact of the suggested approach.

1 Introduction

Europe faces a double urgency to increase energy efﬁciency: on the one hand, caused by the war in

Ukraine, on the other hand, due to the continuous rise in electricity consumption [1]. Tackling the

energy efﬁciency problem through consumers’ behavioral change is an obvious, however challenging

solution. People often need a guidance, and sometimes a soft nudge to put the intentions into actions

[2], for instance, to change the timing of appliances usage. Recommender systems can suggest energy-

efﬁcient actions to facilitate such behavioral change. To increase the trust in the recommendation

system, and, thus, the acceptance rate of recommendations, users need to understand why and how

the model makes its predictions [3], [4]. Thus, the recommendation system should be explainable.

The existing research on explainability in recommender systems for energy-efﬁcient smart homes is

very scarce [5]. [6] provide a thorough literature review on explainability in recommender systems for

other application domains. However, most existing approaches are not applicable to the smart home

area because of the missing data structures. [7] design an explainable context-aware recommendation

system for a smart home ecosystem. They show that displaying the advantages and the reasoning

behind recommendations lead to a 19% increase in acceptance rate. To our best knowledge, the issue

∗Corresponding author.

Tackling Climate Change with Machine Learning: workshop at NeurIPS 2022.

arXiv:2210.11218v2 [cs.MA] 4 Jan 2023

of explainability in multi-agent recommendation systems for energy-efﬁcient smart homes has not

been studied yet.

Our contributions are twofold. First, we suggest an explainable multi-agent recommendation system

for energy efﬁciency in private households. In particular, we extend a novel multi-agent approach

of [8] by designing an Explainability Agent that provides explainable recommendations for optimal

appliance scheduling in a textual and visual manner. Second, we enhance the predictive capacity

of other agents by including weather data and applying state-of-the-art models. We also provide

an overview of predictive and explainability performance. We provide a comprehensive tutorial in

Jupyter Notebook in GitHub 1for all the steps described in this paper and beyond.

2 Explainable multi-agent recommendation system

[8] introduce a utility-based context-aware multi-agent recommendation system that provides load

shifting recommendations for household devices for the next 24 hours. Their system includes six

agents: Price Agent (prepares external hourly electricity prices), Preparation Agent (prepares data

for the other agents), Availability Agent (predicts the hourly user availability for the next 24 hours),

Usage Agent (calculates the devices’ usage probabilities for the prediction day), Load Agent (extracts

the typical devices’ loads), and Recommendation Agent (collects the inputs from the other agents

and provides recommendations). The multi-agent architecture of [8] is ﬂexible and can be easily

integrated into existing smart home systems. However, the cost of the simplicity of the approach (i.e.,

they use Logistic Regression for the availability and usage predictions) is a relatively low prediction

accuracy.

We address the limitations in [8] by enhancing the performance of the Availability and the Usage

Agents. In particular, we apply the K-Nearest-Neighbours (KNN), extreme gradient boosting (XG-

Boost), adaptive boosting (AdaBoost), and Random Forest to predict the availability and usage

probabilities. Furthermore, we use Logistic Regression (Logit) and Explainable Boosting Machines

(EBM, see [9]) as inherently explainable models. We propose including the Explainability Agent in

the system (see Figure 1). The explainability models are divided into local and global, depending

on their capability to explain a particular instance or the entire model. Since we want to help the

user understand a single recommendation, we focus on local approaches. In particular, we apply

post-model approaches LIME (local, interpretable, model-agnostic explanation; [10]) and SHAP

(Shapley additive explanations; [11]) as model-agnostic tools that can explain the predictions of the

chosen classiﬁers.

Figure 1: Architecture of the explainable multi-agent recommendation system.

To create an explanation for a recommendation, the Explainability Agent extracts feature importance

from the explainability models. We design the explanation to include two parts: (i) Usage explanation

- which features lead to the speciﬁc device usage prediction for the day? and (ii) Availability

explanation - which features drive the user availability prediction for the hour? We do not include an

explanation for the Load Agent since we do not consider the extracted typical load proﬁle of every

shiftable device as informative to the users. As a result, the Recommendation Agent recommends the

cheapest starting hour within the hours of user availability for the shiftable devices that are likely to

be used on the prediction day with an explanation in a text and visual form. The system provides

no recommendations if the predictions for the availability and usage probabilities are below the

thresholds.

1https://github.com/Humboldt-WI/Explainable_multi-agent_RecSys

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

ExplainableMulti-AgentRecommendationSystemforEnergy-EfcientDecisionSupportinSmartHomesAlonaZharovaHumboldt-UniversitätzuBerlinBerlin,Germanyalona.zharova@hu-berlin.deAnnikaBoerHumboldt-UniversitätzuBerlinBerlin,Germanyboeranni@hu-berlin.deJuliaKnoblauchHumboldt-UniversitätzuBerlinBerlin,Germanyjul...

展开>> 收起<<

Explainable Multi-Agent Recommendation System for Energy-Efﬁcient Decision Support in Smart Homes.pdf

共9页,预览2页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Explainable Multi-Agent Recommendation System for Energy-Efﬁcient Decision Support in Smart Homes

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: