Abstract:Hybrid intelligence aims to enhance decision-making, problem-solving, and overall system performance by combining the strengths of both, human cognitive abilities and artificial intelligence. With the rise of Large Language Models (LLM), progressively participating as smart agents to accelerate machine learning development, Hybrid Intelligence is becoming an increasingly important topic for effective interaction between humans and machines. This paper presents an approach to leverage Hybrid Intelligence towards sustainable and energy-aware machine learning. When developing machine learning models, final model performance commonly rules the optimization process while the efficiency of the process itself is often neglected. Moreover, in recent times, energy efficiency has become equally crucial due to the significant environmental impact of complex and large-scale computational processes. The contribution of this work covers the interactive inclusion of secondary knowledge sources through Human-in-the-loop (HITL) and LLM agents to stress out and further resolve inefficiencies in the machine learning development process.

What problem does this paper attempt to address?

The problem discussed in this paper is how to achieve sustainable and energy-efficient machine learning through the integration of human intelligence (Hybrid Intelligence). As large language models (LLMs) play an increasingly important role in accelerating the development of machine learning, the effectiveness of human-machine interaction has become a major issue. Research suggests that although the ultimate performance is the main goal in the development process of machine learning models, process efficiency and energy efficiency are often overlooked. In addition, due to the impact of the computing process on the environment, energy efficiency has become equally critical. The paper proposes a method that integrates minor knowledge sources interactively into the machine learning process through Human-in-the-Loop (HITL) and LLM agents to identify and address inefficiencies during the training process. Currently, there is a lack of a framework for accurately measuring energy consumption in machine learning training. Therefore, the paper aims to integrate energy awareness by not only tracking hardware usage but also considering the energy footprint of the entire infrastructure. Related work focuses on enhancing energy awareness or improving interactive visualization in machine learning development, but the combination of the two has not been explored yet. The proposed approach in the paper involves leveraging the HITL system and LLM intelligent agents to improve resource efficiency, reduce energy consumption, and allow users to intervene in the training process through visualization tools to identify and address issues related to data quality, model architecture, and hardware efficiency. The evaluation method will measure the effectiveness of the approach through numerical analysis and user feedback, with a focus on performance improvements and efficiency gains learned from the field of Human Activity Recognition (HAR) applications. The ultimate goal is to build a comprehensive framework that connects intelligent visualization and sustainability approaches to optimize resource utilization and energy efficiency in machine learning.

Leveraging Hybrid Intelligence Towards Sustainable and Energy-Efficient Machine Learning

Empowering Large Language Models in Hybrid Intelligence Systems through Data-Centric Process Models

Improving the Efficiency of Human-in-the-Loop Systems: Adding Artificial to Human Experts

Hybrid Training Approaches for LLMs: Leveraging Real and Synthetic Data to Enhance Model Performance in Domain-Specific Applications

AI, Meet Human: Learning Paradigms for Hybrid Decision Making Systems

Hybrid Heterogeneous Clusters Can Lower the Energy Consumption of LLM Inference Workloads

Human-Machine Learning for Data-Driven Scenarios

Hybrid-augmented Intelligence: Collaboration and Cognition.

The future of human-AI collaboration: a taxonomy of design knowledge for hybrid intelligence systems

Addressing the Challenges and Barriers to the Integration of Machine Learning into Clinical Practice: An Innovative Method to Hybrid Human–Machine Intelligence

Synergizing Machine Learning & Symbolic Methods: A Survey on Hybrid Approaches to Natural Language Processing

Metaheuristics and Large Language Models Join Forces: Towards an Integrated Optimization Approach

Leveraging Large Language Models for Hybrid Workplace Decision Support

Hybrid Active Inference

Enhancing Smart Grid Sustainability: Using Advanced Hybrid Machine Learning Techniques While Considering Multiple Influencing Factors for Imputing Missing Electric Load Data

Optimal energy management strategies for hybrid electric vehicles: A recent survey of machine learning approaches

Learning with Analytical Models

Artificial intelligence, human intelligence and hybrid intelligence based on mutual augmentation

Meta-heuristics and deep learning for energy applications: Review and open research challenges (2018–2023)

An Intelligent Hybrid Machine Learning Model for Sustainable Forecasting of Home Energy Demand and Electricity Price