Warning: This paper has contents which may be offensive, or upsetting however this cannot be avoided owing to the nature of the work. With the rise of online hate speech, automatic detection of Hate Speech, Offensive texts as a natural language processing task is getting popular. However, very little research has been done to detect unintended social bias from these toxic language datasets. This paper introduces a new dataset ToxicBias curated from the existing dataset of Kaggle competition named "Jigsaw Unintended Bias in Toxicity Classification". We aim to detect Figure 1: An illustrative example of ToxicBias. During social biases, their categories, and targeted the annotation process, hate speech/offensive text groups. The dataset contains instances annotated is provided without context. Annotators are asked to for five different bias categories, viz., mark it as biased/neutral and to provide category, target, gender, race/ethnicity, religion, political, and and implication if it has biases.

The cloning of a speaker's voice using an untranscribed reference sample is one of the great advances of modern neural text-to-speech (TTS) methods. Approaches for mimicking the prosody of a transcribed reference audio have also been proposed recently. In this work, we bring these two tasks together for the first time through utterance level normalization in conjunction with an utterance level speaker embedding. We further introduce a lightweight aligner for extracting fine-grained prosodic features, that can be finetuned on individual samples within seconds. We show that it is possible to clone the voice of a speaker as well as the prosody of a spoken reference independently without any degradation in quality and high similarity to both original voice and prosody, as our objective evaluation and human study show. All of our code and trained models are available, alongside static and interactive demos.

Recent findings have shown that highly over-parameterized Neural Networks generalize without pretraining or explicit regularization. It is achieved with zero training error, i.e., complete over-fitting by memorizing the training data. This is surprising, since it is completely against traditional machine learning wisdom. In our empirical study we fortify these findings in the domain of fine-grained image classification. We show that very large Convolutional Neural Networks with millions of weights do learn with only a handful of training samples and without image augmentation, explicit regularization or pretraining. We train the architectures ResNet018, ResNet101 and VGG19 on subsets of the difficult benchmark datasets Caltech101, CUB_200_2011, FGVCAircraft, Flowers102 and StanfordCars with 100 classes and more, perform a comprehensive comparative study and draw implications for the practical application of CNNs. Finally, we show that a randomly initialized VGG19 with 140 million weights learns to distinguish airplanes and motorbikes with up to 95% accuracy using only 20 training samples per class.

While AI algorithms have shown remarkable success in various fields, their lack of transparency hinders their application to real-life tasks. Although explanations targeted at non-experts are necessary for user trust and human-AI collaboration, the majority of explanation methods for AI are focused on developers and expert users. Counterfactual explanations are local explanations that offer users advice on what can be changed in the input for the output of the black-box model to change. Counterfactuals are user-friendly and provide actionable advice for achieving the desired output from the AI system. While extensively researched in supervised learning, there are few methods applying them to reinforcement learning (RL). In this work, we explore the reasons for the underrepresentation of a powerful explanation method in RL. We start by reviewing the current work in counterfactual explanations in supervised learning. Additionally, we explore the differences between counterfactual explanations in supervised learning and RL and identify the main challenges that prevent adoption of methods from supervised in reinforcement learning. Finally, we redefine counterfactuals for RL and propose research directions for implementing counterfactuals in RL.

The dawn of the digital medicine era, ushered in by increasingly powerful embedded systems and Internet of Things (IoT) computing devices, is creating new therapies and biomedical solutions that promise to positively transform our quality of life. However, the digital medicine revolution also creates unforeseen and complex ethical, regulatory, and societal issues. In this article, we reflect on the ethical challenges facing digital medicine. We discuss the perils of ethical oversights in medical devices, and the role of professional codes and regulatory oversight towards the ethical design, deployment, and operation of digital medicine devices that safely and effectively meet the needs of patients. We advocate for an ensemble approach of intensive education, programmable ethical behaviors, and ethical analysis frameworks, to prevent mishaps and sustain ethical innovation, design, and lifecycle management of emerging digital medicine devices.

The explorative and iterative nature of developing and operating machine learning (ML) applications leads to a variety of artifacts, such as datasets, features, models, hyperparameters, metrics, software, configurations, and logs. In order to enable comparability, reproducibility, and traceability of these artifacts across the ML lifecycle steps and iterations, systems and tools have been developed to support their collection, storage, and management. It is often not obvious what precise functional scope such systems offer so that the comparison and the estimation of synergy effects between candidates are quite challenging. In this paper, we aim to give an overview of systems and platforms which support the management of ML lifecycle artifacts. Based on a systematic literature review, we derive assessment criteria and apply them to a representative selection of more than 60 systems and platforms.

A limiting factor for readout fidelity for superconducting qubits is the relaxation of the qubit to the ground state before the time needed for the resonator to reach its final target state. A technique known as excited state promoted (ESP) readout was proposed to reduce this effect and further improve the readout contrast on superconducting hardware. In this work, we use readout data from IBM's five-qubit quantum systems to measure the effectiveness of using deep neural networks, like feedforward neural networks, and various classification algorithms, like k-nearest neighbors, decision trees, and Gaussian naive Bayes, for single-qubit and multi-qubit discrimination. These methods were compared to standardly used linear and quadratic discriminant analysis algorithms based on their qubit-state-assignment fidelity performance, robustness to readout crosstalk, and training time.

Between 2019 and 2022, Greek media attention has been attracted by a rather unusually high number of femicide cases which have been trending for several weeks up to months in the public debate and one of the contributing factors is the feedback loop between traditional media and social media. In this paper we are investigating the use of the term "\{gamma}\u{psion}{\nu}{\alpha}{\iota}\k{appa}{\omicron}\k{appa}{\tau}{\omicron}{\nu}{\iota}{\alpha}" (femicide) in Greek speaking twitter. More specifically, we approach the problem from a stance detection perspective, aiming to automatically identify user position with regards to the feministic semantics of the word. We also discuss findings from an identity analysis perspective and intercorrelations with hate speech that have been identified in the collected corpus of tweets.

Compositionality, the phenomenon where the meaning of a phrase can be derived from its constituent parts, is a hallmark of human language. At the same time, many phrases are non-compositional, carrying a meaning beyond that of each part in isolation. Representing both of these types of phrases is critical for language understanding, but it is an open question whether modern language models (LMs) learn to do so; in this work we examine this question. We first formulate a problem of predicting the LM-internal representations of longer phrases given those of their constituents. We find that the representation of a parent phrase can be predicted with some accuracy given an affine transformation of its children. While we would expect the predictive accuracy to correlate with human judgments of semantic compositionality, we find this is largely not the case, indicating that LMs may not accurately distinguish between compositional and non-compositional phrases. We perform a variety of analyses, shedding light on when different varieties of LMs do and do not generate compositional representations, and discuss implications for future modeling work.

The Automated Speech Recognition (ASR) community experiences a major turning point with the rise of the fully-neural (End-to-End, E2E) approaches. At the same time, the conventional hybrid model remains the standard choice for the practical usage of ASR. According to previous studies, the adoption of E2E ASR in real-world applications was hindered by two main limitations: their ability to generalize on unseen domains and their high operational cost. In this paper, we investigate both above-mentioned drawbacks by performing a comprehensive multi-domain benchmark of several contemporary E2E models and a hybrid baseline. Our experiments demonstrate that E2E models are viable alternatives for the hybrid approach, and even outperform the baseline both in accuracy and in operational efficiency. As a result, our study shows that the generalization and complexity issues are no longer the major obstacle for industrial integration, and draws the community's attention to other potential limitations of the E2E approaches in some specific use-cases.