In musical (jazz) improvisation, musicians that are just starting out can often feel uncomfortable when being put on the spot by their fellow players. However, when a musician is on their own when practising or leisurely playing, this prevents them from listening to fellow musicians. When a musician wants to experience some notion of co-play when they are on their own, computers and musical generative techniques may be a source of help. We study the extent to which music improvisation algorithms can facilitate such interactions by proposing an experimental framework to evaluate and compare these different algorithms. We achieve this by developing MILES ('Mixed-Initiative musicaL interactivE System'), a generic music improvisation system that allows a musician to improvise with various musical improvisation models and facilitates comparative evaluation of these models. MILES makes use of the 'trading fours' paradigm, where two or more musicians exchange four measures of solo material. We conduct experiments with novice and advanced musicians in expert-pupil and peer-to-peer settings that compare differing algorithms, as well as different variations of similar algorithms. These comparisons are based on self-assessed opinions and third-party grading and ordering of recordings, based on improvisational reciprocity and enjoyment. Symbolic music recording analysis further quantifies the interactivity between the musician and the algorithms. With this experimental setup, we are able to track familiarity and enjoyment of using music improvisation algorithms, and compare different iterations of similar music improvisation algorithms.

Machine Learning (ML) systems are increasingly used in high-stakes, socially impactful domains, requiring attention to improve explainability and trust. However, current Requirements Engineering (RE) techniques often fail to address these human-centered qualities. This research investigates how Behavior-Driven Development (BDD) and Goal-Oriented Requirements Engineering (GORE) can improve the identification and visualization of requirements for explainability and trust. We conducted expert interviews and a survey to see how different stakeholders rate certain BDD scenarios and what they think of conceptual GORE models. Our results show that participants value concise, human-readable BDD scenarios and particularly like the GORE framework of i* to understand stakeholder relationships and system behaviors. The other GORE framework, GR4ML, was found to align more with business goals and addresses other stakeholder perspectives less. We conclude that BDD and GORE can improve explainability in ML system development. Future work should refine modeling tools to better integrate ethical and fairness considerations.

Central banks communicate their monetary policy plans to the public through meeting minutes or transcripts. These communications can have immense effects on markets and are often the subjects of studies in the financial literature. The recent advancements in Natural Language Processing have prompted researchers to analyze these communications using Transformer-based Large Language Model (LLM) classifiers. The use of LLMs in finance and other high-stakes domains calls for a high level of trustworthiness and explainability of those models. We focus on Counterfactual Explanations, a form of Explainable AI that explains a model's classification by proposing an alternative to the original input. We use three types of CE generators for LLM classifiers on a recent dataset consisting of sentences taken from FOMC communications to assess the usability of their explanations. We perform three experiments comparing different types of generators, one using a selection of quantitative metrics and two involving human evaluators, including central bank employees. Our findings suggest that non-expert and expert evaluators prefer counterfactual methods that apply minimal changes to the texts; however, the methods we analyze might not handle the domain-specific vocabulary well enough to generate plausible explanations for our task. We discuss shortcomings in the choice of evaluation metrics in the literature on text CE generators and propose refined definitions of the fluency and plausibility qualitative metrics.

Large language models have achieved breakthroughs in many natural language processing tasks. One of their main appeals is the ability to tackle problems that lack sufficient training data to create a dedicated solution. Manga translation is one such task, a still budding and underdeveloped field, that at the same time deals with a problem even more complex than standard translation. One of its biggest challenges is successfully incorporating the visual modality in translation, to resolve ambiguities. Recently, a new type of LLMs -- multimodal LLMs -- has emerged and showed potential in understanding visual symbolism in narrative pieces like memes and comics. In this work, we investigate whether these models could be useful in the field of manga translation, ultimately helping manga authors and publishers make their works available to wider audiences. Specifically, we evaluate a number of methods based on GPT-4-Turbo for text only translation, image informed translation and volume-level translation. We perform both automatic and human evaluation. Moreover, we contribute new evaluation data -- the first parallel Japanese-Polish manga translation benchmark. Our findings show that our proposed methods are able to achieve state of the art results for English, and set a new standard for Polish. We conclude that while this is not a sufficient replacement for a professional human translator, it could help speed up the translation process or be used as a learning aid.

Counterfactual Explanations (CE) are essential for understanding the predictions of black-box models by suggesting minimal changes to input features that would alter the output. Despite their importance in Explainable AI (XAI), there is a lack of standardized metrics to assess the plausibility and faithfulness of these explanations. This paper reviews evaluation procedures in literature and proposes novel formal metrics for evaluating the plausibility and faithfulness of counterfactual explanations, addressing the existing limitations. Plausibility is defined as the coherence of explanations with the true data-generating process, while faithfulness refers to the accuracy of explanations in representing the model's reasoning. We discuss the shortcomings of existing evaluation procedures and metrics for measuring plausibility and faithfulness and consequently compare our proposed metrics with existing ones, highlighting their advantages and disadvantages. The proposed metrics are then empirically validated through experiments across multiple models and datasets, demonstrating their model-agnostic nature and reliability. Our findings indicate that the proposed metrics provide a correct and reliable means to quantify the plausibility and faithfulness of counterfactual explanations, thereby allowing one to gauge their feasibility and trustworthiness consistently.

Counterfactual explanations (CEs) can be used to gain useful insights into the behaviour of opaque classification models, allowing users to make an informed decision when trusting such systems. Assuming the CEs of a model are faithful (they well represent the inner workings of the model), an explainable model generates plausible CEs (i.e. CEs fitting the real-world distribution of the data). This raises the question of whether classifiers explicitly designed to model the distribution of the data, such as energy-based models, are inherently more explainable. This work focuses on the evaluation of joint energy-based models (JEMs) in combination with the Energy-Constrained Conformal Counterfactuals (ECCCo) generator, with the goal of identifying if the generative capability of a model influences its explainability. Since ECCCo has been designed specifically to generate more faithful CEs, it makes it possible to use the CEs plausibility as a proxy of the model explainability. Two experiments have been performed to evaluate the effect of variations of generative capability within the same JEM architecture and the difference between JEMs and classically trained classifiers. Despite the experiments not having established a clear correlation between generative capability and explainability of a model, various research avenues are still open to explore in future works

Adversarial Training has emerged as the most reliable technique to make neural networks robust to gradient-based adversarial perturbations on input data. Besides improving model robustness, preliminary evidence presents an interesting consequence of adversarial training -- increased explainability of model behaviour. Prior work has explored the effects of adversarial training on gradient stability and interpretability, as well as visual explainability of counterfactuals. Our work presents the first quantitative, empirical analysis of the impact of model robustness on model explainability by comparing the plausibility of faithful counterfactuals for both robust and standard networks. We seek to determine whether robust networks learn more plausible decision boundaries and representations of the data than regular models, and whether the strength of the adversary used to train robust models affects their explainability. Our finidngs indicate that robust networks for image data learn more explainable decision boundaries and representations of data than regular models, with more robust models producing more plausible counterfactuals. Robust models for tabular data, however, only conclusively exhibit this phenomenon along decision boundaries and not for its overall data representations, possibly due to its high robustness-accuracy trade-off and the difficulties associated with traditional adversarial training due to its innate properties. We believe our work can help guide future research towards improving the robustness of machine learning models keeping their explainability in mind.

Counterfactual explanations can be applied to algorithmic recourse, which is concerned with helping individuals in the real world overturn undesirable algorithmic decisions. They aim to provide explanations to opaque machine learning models. Not all generated points are equally faithful to the model, nor equally plausible. On the other hand, predictive uncertainty quantification is used to measure the degree of certainty a model has in its predictions. Previously, it has been shown that it is possible to generate more plausible counterfactual explanations utilising predictive uncertainty. This work investigates this further by using multiple models innately supporting uncertainty quantification and comparing the produced counterfactual explanations to those produced by the models' ordinary counter-part. Predictive uncertainty tends to enhance the plausibility of the counterfactuals on visual datasets. Furthermore, we are positive that predictive uncertainty correlates proportionally with plausibility. This correlation has important implications for both research and real-world applications, as it suggests that integrating uncertainty quantification in model development can improve the quality and trustworthiness of algorithmic explanations.

In recent years, the need for explainable artificial intelligence (XAI) has become increasingly important as complex black-box models are used in critical applications. While many methods have been developed to interpret these models, there is also potential in enhancing the models themselves to improve their inherent explainability. This paper investigates various techniques aimed at improving the explainability of black-box models. Through a systematic literature review, these techniques are categorized, and their impact on predictive uncertainty, adversarial robustness, and generative capacity is analyzed to understand how these factors contribute to the overall explainability. The snowballing methodology is used for the systematic literature review, starting with papers retrieved from four databases: IEEExplore, Scopus, ArXiv, and the ACM Digital Library to form the initial set. This process continued with backward and forward snowballing through four iterations, resulting in a total of 50 papers reviewed. Only papers focused on improving model explainability are included in the review. Due to time limitations, additional search constraints are applied for feasibility. The initial set of papers is filtered to those published since 2013. These constraints and their possible impacts are considered when interpreting the results. Findings reveal that techniques such as Bayesian approaches and variational inference, adversarial robustness, model compression and distillation, uncertainty and ensembles, regularization, self-explaining models, and hybrid techniques are used for advancing model explainability. The paper concludes with a discussion on the implications of these techniques for future research.

Counterfactual explanations (CEs) are emerging as a crucial tool in Explainable AI (XAI) for understanding model decisions. This research investigates the impact of various factors on the quality of CEs generated for classification tasks. We explore how inter-class distance, data imbalance, balancing techniques, the presence of biased classifiers, and decision thresholds influence CE quality. To answer these research questions, we conduct experiments on various datasets, classification models and counterfactual generators. The datasets include the MNIST and GMSC dataset. The models include well-established models like MLP and Random Forest, along with the novel NeuroTree model. The generators include the method proposed by Wachter et al. and the REVISE method. We evaluate how different factors affect CE quality by performing an extensive experimental analysis. Our findings demonstrate that increasing inter-class distance degrades CE quality, particularly explanation plausibility. Data imbalance showed minimal impact, while balancing techniques yielded a slight improvement in CE plausibility, especially for the minority class. Classifiers biased towards specific subgroups resulted in lower CE quality for those subgroups. We observed limited evidence for a consistent amplification effect of decision thresholds. This research utilizes various datasets and classification models, including the novel NeuroTree model. Our findings contribute to XAI by providing insights into factors affecting CE quality and highlighting areas for future development, particularly regarding fairness and handling imbalanced data.

Due to climate change, man-made conflicts, and rising inflation, a growing number of people around the world are struggling to have consistent access to safe and nutritious food. This phenomenon is known as food insecurity (FI). Therefore, we take in this thesis the first steps towards developing a monitoring process for assessing FI using Human-Centered AI (HCAI). We developed this process for, and in collaboration with, the Integrated Food Security Phase Classification (IPC). The IPC is an organization that helps countries classify levels of food insecurity in their regions to inform humanitarian decision-making. During our research process, we found that any form of HCAI for the IPC would need to be informed by input from their domain experts, and we concluded that we could not start implementing HCAI until we found a way to formalize their input in a way that was robust and suited their technical capabilities. To this end, we ran an experiment with 18 IPC experts in Malawi to see whether they could quantify their assumptions by setting thresholds for food security drivers. The results are encouraging but show that there is still much to be done to bridge the gap between domain knowledge and technical expertise. We also show in this thesis that there is a lack of real-life case studies on HCAI development and share therefore our lessons learned from our real-world HCAI case study on FI monitoring. In this way, we hope to promote the development of a practice-informed methodology for HCAI.

Recommender systems are widely used to help users navigate vast content catalogs, but they often limit users to suggestions that closely match their existing preferences, creating "filter bubbles" that discourage exploration. We focus on solving this problem in the context of music recommendations, helping users discover and develop new musical tastes. We embed a knowledge graph containing expert-curated metadata, user interaction data, and audio similarity features, into a representation space where similar songs are mapped closely together. This enables the system to gradually guide users from their current preferences toward a new genre through personalized recommendations. Additionally, we apply a Bayesian active learning approach to iteratively update user preference models based on feedback, balancing exploration and exploitation to ensure user satisfaction while gathering information on the user's new preferences. We conducted a user study to evaluate the approach, demonstrating that a gradual, interactive approach outperforms directly introducing users to a new genre, increasing user engagement and their affinity toward the target genre. This research highlights the value of gradual, user-driven exploration in creating better music discovery experiences. Based on our findings, we provide recommendations for industry stakeholders and discuss opportunities for future research on targeted exploration in music recommendation.

The aim of algorithmic recourse (AR) is generally understood to be the provision of "actionable" recommendations to individuals affected by algorithmic decision-making systems in an attempt to present them with the capacity to take actions that would guarantee more desirable outcomes in the future. Over the past few years, the literature has predominantly focused on the development of solutions to generate "actionable" counterfactual explanations that further satisfy various desiderata, such as diversity or robustness. We believe that algorithmic recourse, by its nature, should be seen as a practical challenge: real-world decision-making systems are complex dynamic entities involving various actors – end users, domain experts, system owners, etc. – engaging in social and technical processes. Thus, research on algorithmic recourse should account for the characteristics of systems where such mechanisms could be implemented. This necessitates a rich understanding of the problem space of AR but, as we observe, it remains largely uncharted in the existing literature.

We focus on algorithmic recourse in real-world contexts, applying Design Science Research methods to bridge the gap between its technical affordances and the social constraints of real-world decision-making systems where it could be applied. First, we conduct a systematized literature review of 127 publications to learn about the authors' perception of the problem. Next, we consider a case study of a risk profiling model developed to support the authorities of a major Dutch city in the detection of welfare fraud. We employ a desk research approach to learn about the system, reinforce our understanding of the requirements for algorithms in public administration settings through interviews with experts, and make use of accident analysis methodologies to theorize about the value of AR interventions in this setting. We draw on these insights to propose a conceptual framework for the evaluation of AR in real-world contexts and provide its proof-of-concept instantiation as a simulation tool that facilitates the study of such mechanisms within decision-making processes. Finally, we design and prove an algorithm to generate actionable recommendations in expert systems. These are commonly used in public administration systems but overlooked in existing research.

On the example of our endeavor, we learn about the ways to strengthen the connections between the problem space and the solution space of algorithmic recourse. We argue that AR can be discussed on three levels of complexity: (1) as actionable recommendations, (2) as the process of improving outcomes, or (3) as the task of developing mechanisms to support end-users in this process. We advocate for computer science authors to focus on the final, broadest meaning of the challenge to improve the applicability of their solutions in real-world contexts. We also encourage researchers from other fields to contribute their perspectives and for practitioners to support further research by building upon our approach to reason about the place for AR solutions in their domains of expertise.

The evaluation metrics commonly used for machine learning models often fail to adequately reveal the inner workings of the models, which is particularly necessarily in critical fields like healthcare. Explainable AI techniques, such as counterfactual explanations, offer a way to uncover a model’s internal process. However, these explanations are in literature often used for recourse actions rather than for testing a model’s internal mechanism. In this paper, we propose a proof of concept for a framework which uses counterfactual explanation to evaluate the inner workings of biological machine learning models that use gene expression data. Our approach involves comparing the change of gene expression observed in the original data to the change of gene expression observed between the factual and counterfactual data. The change of gene expression is quantified using the log fold change. Additionally, we expand the definition of faithfulness and introduce a new metric that measures how faithful the generated counterfactual explanations represent the model. This metric should ensure that the explanations accurately reflect the model’s true internal process.

In the task of music style transfer, the symbolic music representation based on Musical Instrument Digital Interface (MIDI) files has always been a popular research medium. By using such representation, some mature models for image style transfer can also be applied to this scenario, such as Cycle-consistent Generative Adversarial Networks (CycleGAN). However, this MIDI-based data representation is not suitable for guitar music because it does not support unique expressive information of guitar playing, such as bending, sliding, or other playing techniques. DadaGP, a dataset made up of guitar-specific format files (tablatures) and their rendered text-like tokens, enables us to perform symbolic guitar music style transfer leveraging expressive guitar playing information, and to produce playable guitar tablatures. We first adopt K-hot encoding to transform the task from sequence generation to binary classification of multiple variables, and use top-$k$ sampling to reproduce sequences from output K-hot vectors. We then propose a novel model we call CycleGMT, a CycleGAN-based model for symbolic guitar music style transfer. Finally, to mitigate the severe sparsity in the data and its resulting content loss, we adopt a skip connection between the input and output of the model, successfully achieving style-transferred music whose quality being competitive with human-composed remixes, while the musical complexity of the style-transferred music can be controlled by adjusting the value of $k$ in top-$k$ sampling.

Machine Learning models are nowadays infused into all aspects of our lives. Perhaps one of its most common applications regards recommender systems, as they facilitate users' decision-making processes in various scenarios (e.g., e-commerce, social media, news, online learning, etc.). Training performed on large volumes of data is what ultimately drives such a system to provide meaningful recommendations, and yet there has been observed a lack of standardized practices when it comes to data collection and annotation methods for Machine Learning datasets. This research paper systematically identifies and synthesizes such processes by examining existing literature on recommender systems. The review includes 100 most-cited papers from the most impactful venues within the Computing and Information Technology field. Multiple facets of the employed techniques are touched upon, such as reported human annotations and annotator diversity, label quality, and the public availability of training datasets.
Recurrent use of just a few benchmark datasets, poor documentation practices, and reproducibility issues in experiments are some of the most striking findings uncovered by this study. A discussion is centered around the necessity of transitioning from reliance solely on algorithmic performance metrics in favor of prioritizing data quality and fit. Finally, valid concerns are raised when it comes to biases and socio-psychological factors inherent in the datasets, and further exploration of embedding these early in the design of ML models is suggested.

This systematic review investigates the practices and implications of human annotations in machine learning (ML) research. Analyzing a selection of 100 papers from the IEEE Access Journal, the study explores the data collection and reporting methods employed. The findings reveal a prevalent lack of standardization and formalization in the annotation process. Key details such as annotation sources, number of annotators, and formal instructions are frequently neglected, possibly compromising the quality and effectiveness of ML algorithms. Domain-specific implications are discussed, highlighting the need for comprehensive annotation practices in areas like medical diagnostics, language processing, and intelligent vehicle systems. The study contributes to the field by emphasizing the importance of standardized procedures and transparency in ML research. Future research is recommended to develop systematic annotation methodologies and examine the impact of subpar annotation on data quality.

Depression diagnosis and treatment remain difficult tasks that could be improved with machine learning models. But those automatic systems should be reliable to apply in clinical psychology settings. Performing predictions in this field is most commonly done using supervised learning models, which rely on well-established annotations. Therefore this paper examines annotation practices in research surrounding depression and provides a perspective on the quality of established methods. Firstly, 80 papers were surveyed in terms of reported annotation practices. Then the results were collected and analyzed. The findings suggest that papers from the Computer Science domain would benefit from the utilization of expert knowledge and better practices of human verification. While papers across all domains missed important information on the annotation process and rarely (20\% of papers) provided input data.

In the machine learning research community, significant importance is given to the optimization of techniques which are employed once a benchmark dataset is given. However, less importance is assigned to the quality of these datasets and to how these datasets are obtained. In this work, we look into annotation practices in the research area of affective computing, analysing datasets of emotion classification tasks from text, video, audio, EEG data and more. We find annotation practices of varying quality and recommend that annotation practices be improved, especially with regard to multiple annotator overlap.

This research examines transparency between ICASSP conference papers and the dataset documentations related to the datasets' annotation practices. Top-cited 5 papers and 51 unique resources in total were considered. All of the selected papers utilized at least one dataset. For every paper, an extensive metadata search has done to reach the initial datasource of those datasets. These searches happened both within the paper contents such as sections and references along with outside the paper contents through the way of extensive web queries. Analysis of the papers published from 2021 and 2022 and their relevant datasets revealed varying levels of transparency. Original dataset creators provide comprehensive information, while papers using modified datasets offer limited details on initial annotations. Emphasizing the need for accountability, this study suggests that papers utilizing datasets should trace back to the initial dataset and provide explicit comments. The findings underscore the importance of ensuring sufficient information in initial datasets and promoting transparency and traceability in dataset annotation practices within the ICASSP community.