The Electronics 2023 Best Paper Award has been granted to the following paper:

“Evaluating the Quality of Machine Learning Explanations: A Survey on Methods and Metrics”
by Jianlong Zhou, Amir H. Gandomi, Fang Chen and Andreas Holzinger
Electronics 2021, 10(5), 593; https://doi.org/10.3390/electronics10050593

The winners will receive CHF 500 and a chance to publish a paper free of charge after peer review in Electronics (ISSN: 2079-9292) in 2023.

 Dr. Jianlong Zhou is an Associate Professor at the School of Computer Science, Faculty of Engineering and IT, University of Technology Sydney, Australia, leading the UTS Human-Centred AI research lab. His current work focuses on AI for social good, AI fairness, AI explainability, smart agriculture, visual analytics, behaviour analytics, human-computer interaction, and related applications.

The following is a short Q&A with Dr. Jianlong Zhou:

1. Could you please briefly introduce the main content of the winning paper?
The content of this paper is a literature review on the evaluation of the explanation quality in machine learning, and we carried out a deep literature review on different approaches to evaluate the quality of explanations, which include qualitative and quantitative research.

2. Could you describe the difficulties and breakthrough innovations in this research field?
The most difficult part of this article, in my opinion, is the design. In the beginning, there are many different approaches, and it's a mess. Somebody has treated this topic in this way. Somebody has treated this issue in a different way. And there are other people in different areas. For example, a social scientist, even a psychologist, will also have some discussions about this.

3. How was your experience submitting to Electronics?
My experience with Electronics is that it is very impressive, very good, and I can get feedback very quickly, which I like very much, and I plan to submit my work in the future.

4. What were some of the biggest challenges you faced?
We had to think about a very good idea for a framework to frame this literal view. How do we categorize this work or existing work, and how do we find interesting ideas or challenges in such a review. So, we finally divided the work into a qualitative and quantitative approach, and qualitative is also further divided into different areas to those in the quantitative approach. In each country, there are many explanations for machine learning itself. There are also many different categories, for example, local explanation and global explanation, for each explanation. Categories are also different methods. However, we don't know whether their explanation is a real explanation or not. This kind of work is very challenging, very difficult. And why? Because most of the time we don't have a ground truth, we don't have a ground truth for the explanation. So, it's very difficult work. They only state that they have developed a new explanatory method, but they do not evaluate the quality. They have evaluated some papers. They use some simulated data. They use very simple cases. For example, they just say that they know the ground truth of the explanation and just want to measure the weather. The explanations they generate are closer to the baseline data. For our literature review, we categorized these papers and divided them into qualitative and quantitative. We also found that the evaluation of explanations is more human-focused. We also reviewed a lot of work on human-centered assessment. The evaluation of the quality of machine-learning explanations belonged to two areas: the human-computer interaction domain and the machine-learning domain.

5. How do you think open access impacts authors?
I think that open access is challenging and helps to improve the impact of publications. Our work has to do with electronics, which fits within the scope of your journal, and when readers search for similar topics, they can easily download the full PDF of the article. This way, our work could be easily cited or read, increasing the impact of the work on the research community.

We want to congratulate Dr. Jianlong Zhou for winning the Electronics 2023 Best Paper Award.