원문정보
초록
영어
It is difficult for non-experts to build machine learning (ML) models at the level that satisfies their needs. Deep learning models are even more challenging because it is unclear how to improve the model, and a trial-and-error approach is not feasible since training these models are time-consuming. To assist these novice users, we examined how interactive and explainable feedback while training a deep learning network can contribute to model performance and users’ satisfaction, focusing on the data preparation process. We conducted a user study with 31 participants without expertise, where they were asked to improve the accuracy of a deep learning model, varying feedback conditions. While no significant performance gain was observed, we identified potential barriers during the process and found that interactive and explainable feedback provide complementary benefits for improving users’ understanding of ML. We conclude with implications for designing an interface for building ML models for novice users.
목차
1. Introduction
2. Related Work
2.1 Interactive Feedback for Building ML Models
2.2 Explaining the Performance of ML Models
3. User Study
3.1 Experimental Conditions
3.2 Participants
3.3 Apparatus
3.4 Procedure
4. Findings
4.1 The Impacts of Feedback Conditions on Model Accuracy
4.2 The Understanding of Building a Better ML Model
4.3 The Use and Subjective Assessments of Feedback Features
4.4 Perceived Task Loads and Observed Behaviors of Novices
5. Discussion
5.1 Discrepancy Between Understanding of ML and Accuracy
5.2 Understanding of the Volume and Variety of Training Data
5.3 Trade-offs Between Feedback Types
5.4 Non-Experts’ Misconceptions of Machine Learning Models
5.5 Risk of Providing Incomplete Feedback to Novice Users
5.6 Feasibility of Providing Feedback for Educational Purpose
6. Conclusion and Future Work
References