While multiple emotional speech corpora exist for commonly spoken languages, there is a lack of functional datasets for smaller (spoken) languages, such as Danish. To our knowledge, Danish Emotional Speech (DES), published in 1997, is the only other database of Danish emotional speech. We present EmoTale1; a corpus comprising Danish and English speech recordings with their associated enacted emotion annotations. We demonstrate the validity of the dataset by investigating and presenting its predictive power using speech emotion recognition (SER) models. We develop SER models for EmoTale and the reference datasets using self-supervised speech model (SSLM) embeddings and the openSMILE feature extractor. We find the embeddings superior to the hand-crafted features. The best model achieves an unweighted average recall (UAR) of 64.1% on the EmoTale corpus using leave-one-speaker-out cross-validation, comparable to the performance on DES.1Link to the dataset and source code:https://github.com/snehadas/EmoTale

We introduce EmoLIME1, a version of local interpretable model-agnostic explanations (LIME) for black-box Speech Emotion Recognition (SER) models. To the best of our knowledge, this is the first attempt to apply LIME in SER. EmoLIME generates high-level interpretable explanations and identifies which specific frequency ranges are most influential in determining emotional states. The approach aids in interpreting complex, high-dimensional embeddings such as those generated by end-to-end speech models. We evaluate EmoLIME, qualitatively, quantitatively, and statistically, across three emotional speech datasets, using classifiers trained on both hand-crafted acoustic features and Wav2Vec 2.0 embeddings. We find that EmoLIME exhibits stronger robustness across different models than across datasets with distribution shifts, highlighting its potential for more consistent explanations in SER tasks within a dataset.

Early diagnosis and intervention are crucial for mild cognitive impairment (MCI), as MCI often progresses to more severe neurodegenerative conditions. In this study, we explore utilizing deep learning for MCI detection without loosing the interpretability provided by feature-based approaches. We used a dataset consisting of 90 MCI patients and 91 controls collected via a remote assessment platform and analyzed the participants' spontaneous speech responses to the Patient Report of Problems (PROP) which asks patients to report their most bothersome general health problems. The proposed deep neural network, which features a bottleneck layer including 13 interpretable symptom domains, achieved an AUC of 0.62, thereby outperforming a set of feature-based classifiers while ensuring interpretability due to the bottleneck layer. We further illustrated the model's interpretability by examining how the predicted PROP domains influence final predictions using Shapley values.

Prototypical self-explainable classifiers have emerged to meet the growing demand for interpretable AI systems. These classifiers are designed to incorporate high transparency in their decisions by basing inference on similarity with learned prototypical objects. While these models are designed with diversity in mind, the learned prototypes often do not sufficiently represent all aspects of the input distribution, particularly those in low density regions. Such lack of sufficient data representation, known as representation bias, has been associated with various detrimental properties related to machine learning diversity and fairness. In light of this, we introduce pantypes, a new family of prototypical objects designed to capture the full diversity of the input distribution through a sparse set of objects. We show that pantypes can empower prototypical self-explainable models by occupying divergent regions of the latent space and thus fostering high diversity, interpretability and fairness.

Modeling non-stationary data is a challenging problem in the field of continual learning, and data distribution shifts may result in negative consequences on the performance of a machine learning model. Classic learning tools are often vulnerable to perturbations of the input covariates, and are sensitive to outliers and noise, and some tools are based on rigid algebraic assumptions. Distribution shifts are frequently occurring due to changes in raw materials for production, seasonality, a different user base, or even adversarial attacks. Therefore, there is a need for more effective distribution shift detection techniques.In this work, we propose a continual learning framework for monitoring and detecting distribution changes. We explore the problem in a latent space generated by a bio-inspired self-organizing clustering and statistical aspects of the latent space. In particular, we investigate the projections made by two topology-preserving maps: the Self-Organizing Map and the Scale Invariant Map. Our method can be applied in both a supervised and an unsupervised context. We construct the assessment of changes in the data distribution as a comparison of Gaussian signals, making the proposed method fast and robust. We compare it to other unsupervised techniques, specifically Principal Component Analysis (PCA) and Kernel-PCA. Our comparison involves conducting experiments using sequences of images (based on MNIST and injected shifts with adversarial samples), chemical sensor measurements, and the environmental variable related to ozone levels. The empirical study reveals the potential of the proposed approach.

Speech enhancement techniques improve the quality or the intelligibility of an audio signal by removing unwanted noise. It is used as preprocessing in numerous applications such as speech recognition, hearing aids, broadcasting and telephony. The evaluation of such algorithms often relies on reference-based objective metrics that are shown to correlate poorly with human perception. In order to evaluate audio quality as perceived by human observers it is thus fundamental to resort to subjective quality assessment and in doing so we identify subgroups of users where the subjective assessments correlate better to objective metrics. In this paper, a user evaluation based on crowdsourcing (subjective) and the Comparison Category Rating (CCR) method is compared against the DNS-MOS, ViSQOL and 3QUEST (objective) metrics. The overall quality scores of three speech enhancement algorithms from real time communications (RTC) are used in the comparison using the P.808 toolkit. Results indicate that while the CCR scale allows participants to identify differences between processed and unprocessed audio samples, two groups of preferences emerge: some users rate positively by focusing on noise suppression processing, while others rate negatively by focusing mainly on speech quality. We further present results on the parameters, size considerations and speaker variations that are critical and should be considered when designing the CCR-based crowdsourcing evaluation1.

In recent years, speech emotion recognition (SER) has been used in wide ranging applications, from healthcare to the commercial sector. In addition to signal processing approaches, methods for SER now also use deep learning techniques which provide transfer learning possibilities. However, generalizing over languages, corpora and recording conditions is still an open challenge. In this work we address this gap by exploring loss functions that aid in transferability, specifically to non-tonal languages. We propose a variational autoencoder (VAE) with KL annealing and a semi-supervised VAE to obtain more consistent latent embedding distributions across data sets. To ensure transferability, the distribution of the latent embedding should be similar across non-tonal languages (data sets). We start by presenting a low-complexity SER based on a denoising-autoencoder, which achieves an unweighted classification accuracy of over 52.09% for four-class emotion classification. This performance is comparable to that of similar baseline methods. Following this, we employ a VAE, the semi-supervised VAE and the VAE with KL annealing to obtain a more regularized latent space. We show that while the DAE has the highest classification accuracy among the methods, the semi-supervised VAE has a comparable classification accuracy and a more consistent latent embedding distribution over data sets.1

We propose a novel method to co-train deep convolutional neural networks for data sets of differing position specific data. This is an advantage in chemometrics where individual measurements represent exact chemical compounds, e.g. for given wavelengths, and thus signals cannot be translated or resized without disturbing their interpretation. Our approach outperforms transfer learning for three small data sets co-trained with a medium sized data set.

Course evaluations are widely used by educational institutions to assess the quality of teaching. At the course evaluations, students are usually asked to rate different aspects of the course and of the teaching. We propose to apply canonical correlation analysis (CCA) in order to investigate the degree of association between how students evaluate the course and how students evaluate the teacher. Additionally it is possible to reveal the structure of this association. Student evaluations data is characterized by high correlations between the variables within each set of variables, therefore two modifications of the CCA method; regularized CCA and sparse CCA, together with classical CCA were applied to find the most interpretable model. Both methods give results with increased interpretability over traditional CCA on the present student evaluation data. The method shows robustness when evaluations over several years are examined.

Universities have varying policies on how and when to perform student evaluations of courses and teachers. More empirical evidence of the consequences of such policies on quality enhancement of teaching and learning is needed. A study (35 courses at the Technical University of Denmark) was performed to illustrate the effects caused by different handling of mid-term course evaluations on student's satisfaction as measured by end-of-term evaluations. Midterm and end-of-term course evaluations were carried out in all courses. Half of the courses were allowed access to the midterm results. The evaluations generally showed positive improvements over the semester for courses with access, and negative improvements for those without access. Improvements related to: Student learning, student satisfaction, teaching activities, and communication showed statistically significant average differences of 0.1-0.2 points between the two groups. These differences are relatively large compared to the standard deviation of the scores when student effect is removed (approximately 0.7). We conclude that university policies on course evaluations seem to have an impact on the development of the teaching and learning quality as perceived by the students and discuss the findings.

Extensive research has been done on student evaluations of teachers and courses based on quantitative data from evaluation questionnaires, but little research has examined students' written responses to open-ended questions and their relationships with quantitative scores. This paper analyzes such kind of relationship of a well established course at the Technical University of Denmark using statistical methods. Keyphrase extraction tool was used to find the main topics of students' comments, based on which the qualitative feedback was transformed into quantitative data for further statistical analysis. Application of factor analysis helped to reveal the important issues and the structure of the data hidden in the students' written comments, while regression analysis showed that some of the revealed factors have a significant impact on how students rate a course.

UNLABELLED: NetRaVE is a small suite of R functions for generating dependency networks using sparse regression methods. Such networks provide an alternative to interpreting 'top n lists' of genes arising out of an analysis of microarray data, and they provide a means of organizing and visualizing the resulting information in a manner that may suggest relationships between genes. AVAILABILITY: NetRaVE is freely available for academic use and has been tested in R 2.10.1 under Windows XP, Linux and Mac OS X. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.