Responsible and Representative Multimodal Data Acquisition and Analysis: On Auditability, Benchmarking, Confidence, Data-Reliance & Explainability
arXiv:1903.07171
Abstract
The ethical decisions behind the acquisition and analysis of audio, video or physiological human data, harnessed for (deep) machine learning algorithms, is an increasing concern for the Artificial Intelligence (AI) community. In this regard, herein we highlight the growing need for responsible, and representative data collection and analysis, through a discussion of modality diversification. Factors such as Auditability, Benchmarking, Confidence, Data-reliance, and Explainability (ABCDE), have been touched upon within the machine learning community, and here we lay out these ABCDE sub-categories in relation to the acquisition and analysis of multimodal data, to weave through the high priority ethical concerns currently under discussion for AI. To this end, we propose how these five subcategories can be included in early planning of such acquisition paradigms.
arXiv comment4 pages
References in corpus (6)
- Explainable Artificial Intelligence: Understanding, Visualizing and Interpreting Deep Learning Models
- Multimodal Machine Learning: A Survey and Taxonomy
- P-values: misunderstood and misused
- A survey on measuring indirect discrimination in machine learning
- Deep Learning for Fatigue Estimation on the Basis of Multimodal Human-Machine Interactions
- Gatekeeping Algorithms with Human Ethical Bias: The ethics of algorithms in archives, libraries and society