Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Federated Learning of Socially Appropriate Agent Behaviours in Simulated Home Environments

Saksham Checker, Nikhil Churamani, Hatice Gunes

TL;DR

A novel Federated Continual Learning benchmark that adapts FL-based methods to use state-of-the-art Continual Learning methods to continually learn socially appropriate agent behaviours under different contextual settings.

Abstract

As social robots become increasingly integrated into daily life, ensuring their behaviours align with social norms is crucial. For their widespread open-world application, it is important to explore Federated Learning (FL) settings where individual robots can learn about their unique environments while also learning from each others' experiences. In this paper, we present a novel FL benchmark that evaluates different strategies, using multi-label regression objectives, where each client individually learns to predict the social appropriateness of different robot actions while also sharing their learning with others. Furthermore, splitting the training data by different contexts such that each client incrementally learns across contexts, we present a novel Federated Continual Learning (FCL) benchmark that adapts FL-based methods to use state-of-the-art Continual Learning (CL) methods to continually learn socially appropriate agent behaviours under different contextual settings. Federated Averaging (FedAvg) of weights emerges as a robust FL strategy while rehearsal-based FCL enables incrementally learning the social appropriateness of robot actions, across contextual splits.

Federated Learning of Socially Appropriate Agent Behaviours in Simulated Home Environments

TL;DR

A novel Federated Continual Learning benchmark that adapts FL-based methods to use state-of-the-art Continual Learning methods to continually learn socially appropriate agent behaviours under different contextual settings.

Abstract

As social robots become increasingly integrated into daily life, ensuring their behaviours align with social norms is crucial. For their widespread open-world application, it is important to explore Federated Learning (FL) settings where individual robots can learn about their unique environments while also learning from each others' experiences. In this paper, we present a novel FL benchmark that evaluates different strategies, using multi-label regression objectives, where each client individually learns to predict the social appropriateness of different robot actions while also sharing their learning with others. Furthermore, splitting the training data by different contexts such that each client incrementally learns across contexts, we present a novel Federated Continual Learning (FCL) benchmark that adapts FL-based methods to use state-of-the-art Continual Learning (CL) methods to continually learn socially appropriate agent behaviours under different contextual settings. Federated Averaging (FedAvg) of weights emerges as a robust FL strategy while rehearsal-based FCL enables incrementally learning the social appropriateness of robot actions, across contextual splits.
Paper Structure (24 sections, 2 figures, 2 tables)

This paper contains 24 sections, 2 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. Learning Scenario: MANNERS-DB Dataset
  4. Experimentation Settings
  5. Input Features and Data Augmentation:
  6. Implementation Details:
  7. Evaluation Metrics:
  8. The FL Benchmark
  9. Compared approaches
  10. FedAvg:
  11. FedBN:
  12. FedProx:
  13. FedOpt:
  14. FedDistill:
  15. Results and Discussion
  16. ...and 9 more sections

Figures (2)

  • Figure 1: Centralised Learning (left) requires robots to share data with the server to train a shared model. FL (middle) allows local model weights or gradients to be aggregated on the server without sharing data. FCL (right) further allows individual robots to incrementally learn tasks, sharing model updates with each other.
  • Figure 2: MANNERS-DB: A living room scenario with the Pepper robot. Adapted from tjomsland2022mind.