Publisher Copyright: © 2026 Owner/Author.

The ever-growing speed at which data are generated nowadays, together with the substantial cost of labeling processes cause Machine Learning models to face scenarios in which data are partially labeled. The extreme case where such a supervision is indefinitely unavailable is referred to as extreme verification latency. On the other hand, in streaming setups data flows are affected by exogenous factors that yield non-stationarities in the patterns (concept drift), compelling models learned incrementally from the data streams to adapt their modeled knowledge to the concepts within the stream. In this work we address the casuistry in which these two conditions occur together, by which adaptation mechanisms to accommodate drifts within the stream are challenged by the lack of supervision, requiring further mechanisms to track the evolution of concepts in the absence of verification. To this end we propose a novel approach, AiGAS-dEVL (Adaptive Incremental neural GAS model for drifting Streams under Extreme Verification Latency), which relies on growing neural gas to characterize the distributions of all concepts detected within the stream over time. Our approach exposes that the online analysis of the behavior of these prototypical points over time facilitates the definition of the evolution of concepts in the feature space, the detection of changes in their behavior, and the design of adaptation policies to mitigate the effect of such changes in the model. We assess the performance of AiGAS-dEVL over several synthetic datasets, comparing it to that of state-of-the-art approaches proposed in the recent past to tackle this stream learning setup. Our results reveal that AiGAS-dEVL performs competitively with respect to the rest of baselines, exhibiting a superior adaptability over several datasets in the benchmark while ensuring a simple and interpretable instance-based adaptation strategy.

Traditional content moderation, while effective in reducing toxicity through content removal or censoring, can discourage user participation by making them feel restricted or unfairly targeted, especially in nuanced discussions. Text detoxification offers a more constructive alternative by rephrasing offensive language into respectful forms. We propose R3-Detox, a Reflect-Reason-Rephrase framework that structures detoxification into three steps within a single prompt. The model identifies potentially toxic elements guided by Shapley values to reduce fabricated predictions, evaluates overall toxicity, and then revises the text to eliminate toxicity while retaining meaning. We augment three offensive text paraphrasing datasets (ParaDetox, Parallel Detoxification, APPDIA) with explicit detoxification reasoning. Evaluated with in-context learning, R3-Detox outperforms state-of-the-art methods, including instruction following models.

Home security and surveillance systems are rapidly evolving, with Artificial Intelligence (AI) playing a transformative role in enhancing safety and threat detection. While several AI methods and datasets for intruder-related risk assessment exist, they predominantly focus on face detection and recognition, leaving a significant gap in addressing high-risk scenarios involving malicious intent, such as theft or harm. The lack of dedicated datasets for recognizing complex intruder activities, such as carrying weapons or engaging in destructive actions like kicking doors or breaking locks, limits the development of robust solutions. This work bridges this gap by introducing the Intruder Activity Recognition Dataset (IARD), a video dataset specifically designed to recognize four critical intruder activities: Armed Intruder, Door Kick, Intruder Inside and Lock Breaking. Leveraging IARD, we thoroughly benchmark various state-of-the-art methods, among which a Vision Transformer is found to achieve an impressive 93.3% accuracy in recognizing intruder actions. Our contribution highlights the potential of IARD in advancing AI-driven surveillance systems, providing a foundational dataset and benchmark for recognizing complex intruder activities.

Deep learning models like chatGPT exemplify AI success but necessitate a deeper understanding of trust in critical sectors. Trust can be achieved using counterfactual explanations, which is how humans become familiar with unknown processes; by understanding the hypothetical input circumstances under which the output changes. We argue that the generation of counterfactual explanations requires several aspects of the generated counterfactual instances, not just their counterfactual ability. We present a framework for generating counterfactual explanations that formulate its goal as a multiobjective optimization problem balancing three objectives: plausibility; the intensity of changes; and adversarial power. We use a generative adversarial network to model the distribution of the input, along with a multiobjective counterfactual discovery solver balancing these objectives. We demonstrate the usefulness of six classification tasks with image and 3D data confirming with evidence the existence of a trade-off between the objectives, the consistency of the produced counterfactual explanations with human knowledge, and the capability of the framework to unveil the existence of concept-based biases and misrepresented attributes in the input domain of the audited model. Our pioneering effort shall inspire further work on the generation of plausible counterfactual explanations in real-world scenarios where attribute-/concept-based annotations are available for the domain under analysis.