Efflux pumps are essential components of cellular detoxification mechanisms, regulating the intracellular accumulation of xenobiotics and endogenous compounds. Among them, P-glycoprotein (P-gp) plays a role in protecting the brain from potentially toxic molecules, and alterations in its function have been associated with neurodegenerative disorders, including Parkinson’s disease (PD). Although P-gp inhibitors have been extensively investigated in the context of multidrug resistance, their effects on neuronal cells remain poorly characterized. In the present study, we investigated the biological effects of the pyridoquinoxaline-based efflux pump inhibitor 2,2′-(pyrido[2,3-g]quinoxaline-2,3-diylbis(methylene))bis(oxy)bis(N-phenylbenzamide) (10d) in PC12 cells, a widely used dopaminergic neuronal model. The effects of 10d were evaluated by an MTT-based cell viability assay, while intracellular and extracellular dopamine (DA) levels and DA metabolites were quantified by high-performance liquid chromatography (HPLC). In addition, the ability of 10d to modulate MPTP-induced neurotoxicity was assessed, alone and in combination with amantadine (AMA), a known antiparkinsonian drug. Exposure to 10d (5 and 10 μM) reduced PC12 cell viability and markedly enhanced MPTP-induced cytotoxicity. Furthermore, 10d altered dopaminergic homeostasis by decreasing intracellular DA levels and modifying DA metabolite profiles, with more pronounced effects following co-treatment with MPTP. The combined administration of 10d, MPTP and AMA produced a stronger disruption of DA metabolism compared with individual treatments or 10d/MPTP co-exposure.

A small library of 23 pyrrole-based tricyclic derivatives bearing bulky amine moieties was synthesized, and all were evaluated for their antiviral activities against ZIKV and SARS-CoV. Three compounds, derivatives 2g, 2h and 2j, elicited interesting activity against ZIKV: compound 2g, containing a bornylamine residue, showed the best activity against Huh-7 cells with EC50 and CC50 values of 0.4 μM and 230.5 μM, respectively, and a Selectivity Index (SI) of 501. All three compounds reduce ZIKV yield primarily by impairing viral protein.

Online continual learning (OCL) methods adapt to changing environments without forgetting past knowledge. Similarly, online time series forecasting (OTSF) is a real-world problem where data evolve in time and success depends on both rapid adaptation and long-term memory. Indeed, time-varying and regime-switching forecasting models have been extensively studied, offering a strong justification for the use of OCL in these settings. Building on recent work that applies OCL to OTSF, this paper aims to strengthen the theoretical and practical connections between time series methods and OCL. First, we reframe neural network optimization as a parameter filtering problem, showing that natural gradient descent is a score-driven method and proving its information-theoretic optimality. Then, we show that using a Student's t likelihood in addition to natural gradient induces a bounded update, which improves robustness to outliers. Finally, we introduce Natural Score-driven Replay (NatSR), which combines our robust optimizer with a replay buffer and a dynamic scale heuristic that improves fast adaptation at regime drifts. Empirical results demonstrate that NatSR achieves stronger forecasting performance than more complex state-of-the-art methods.

We study Evolution Strategies (ES) for continual control, where agents must adapt to changing tasks without forgetting previous ones. On sequential MuJoCo locomotion tasks, naive ES suffers from severe catastrophic forgetting. Replay substantially improves retention and can induce positive transfer, while larger replay budgets reduce plasticity. Overall, these results show that ES can support continual adaptation in control and that replay is an effective mechanism for mitigating forgetting.

Parameter-Efficient Fine-tuned (PEFT) models are frequently downloaded from open repositories by practitioners. This widespread practice creates a significant attack surface, as malicious actors can publish backdoored models that induce specific behaviors in response to predefined triggers. We study the problem of weight-space backdoor detection, where a detector classifier predicts whether a model is malicious using only its weights, enabling a lightweight safety mechanism. Most existing methods are designed and evaluated in a closed-world setting, where the detector is trained and tested on the same attack type. In contrast, we evaluate backdoor detection under novel conditions, including previously unseen attacks and datasets. We propose Z-PEFT, a lightweight meta-classifier that relies exclusively on layer-wise spectral measures for classification. Our experiments show that strong performance in the closed-world setting does not necessarily translate to high accuracy in zero-shot backdoor detection. Among weight-space detectors, Z-PEFT achieves the best performance while maintaining low and scalable computational cost.

Foundation models have transformed machine learning through large-scale pretraining, massive parameterization, and increased test-time compute. Despite surpassing human performance in several domains, these models remain fundamentally limited in continuous operation, experience accumulation, and personalization, capabilities that are central to adaptive intelligence. While continual learning research has long targeted these goals, its historical focus on in-weight learning, i.e., updating a single model’s parameters to absorb new knowledge, has rendered catastrophic forgetting a persistent challenge. We outline a conceptual framework for modular memory-centric architectures that leverage ICL for rapid adaptation and knowledge accumulation, and IWL for stable updates to model capabilities, thereby mitigating catastrophic forgetting and charting a practical roadmap toward continually learning agents.

Introduction: Despite the impressive progress carried out in the field of biomedical sciences in recent decades, the incidence of emerging and neglected lethal viral infections mainly belonging to the Coronaviridae, Filoviridae, Arenaviridae, Bunyaviridae, and Paramyxoviridae families has considerably impaired human health. The worldwide vaccination campaign at the end of the 1970s determined the eradication of smallpox. However, the growing number of cases of diseases linked to orthopoxvirus diseases, such as the recent epidemic of monkeypox zoonosis in various countries around the world, has increased the need for knowledge of these viral pathogens. To date, there is no specific treatement for Monkeypox virus (MPXV) infection. However, several antiviral drugs used to treat Smallpox and other viral infections could also be beneficial for Monkeypox disease. In this study we report the design and synthesis of new, variously substituted benzimidazole derivatives and the evaluation of their cytotoxicity and antiviral activity against representatives of the Orthopoxvirus genus, Vaccinia Virus (VV), closely related to variola virus and MPXV. Methods: A combination of cell-based assays and experimental techniques was used to investigate the cytotoxicity, antiviral activity, and mechanisms of action of the most interesting compound. Results: In our study, new, variously substituted benzimidazoles showed interesting EC50 values against vaccinia and MPXV and a cytotoxic profile in the high micromolar range. Conclusions: Our work shows that the new tested benzimidazole derivatives possess appealing activity and selectivity, accompanied by low cytotoxicity. These results set a valid foundation with which to identify potent and selective anti-Poxvirus agents.

Continual learning (CL) refers to the ability of a model to learn from non-stationary data incrementally. Bayesian continual learning methods are increasingly popular, as they naturally support continual learning through online posterior updates. However, many such methods rely on strong assumptions about data drifts, limiting their applicability to real-world data streams. In this paper, we first define a framework for CL based on mixture models to highlight the limitations of existing Bayesian CL methods. Then, we propose the Class-Domain Infinite Mixture Model (CD-IMM), a non-parametric Bayesian method that adapts dynamically to increasing data complexity over time. We evaluate CD-IMM on mixed class-domain continual learning scenarios involving both class repetitions and novel domains, and compare it to baselines designed for continual learning with pretrained models. Experimental results show that CD-IMM outperforms strong Bayesian continual learning methods.

Low Rank Adaptation (LoRA) is the de facto fine-tuning strategy to generate personalized images from pre-trained diffusion models. Choosing a good rank is extremely critical, since it trades off performance and memory consumption, but today the decision is often left to the community's consensus, regardless of the personalized subject's complexity. The reason is evident: the cost of selecting a good rank for each LoRA component is combinatorial, so we opt for practical shortcuts such as fixing the same rank for all components. In this paper, we take a first step to overcome this challenge. Inspired by variational methods that learn an adaptive width of neural networks, we let the ranks of each layer freely adapt during fine-tuning on a subject. We achieve it by imposing an ordering of importance on the rank's positions, effectively encouraging the creation of higher ranks when strictly needed. Qualitatively and quantitatively, our approach, LoRA^2, achieves a competitive trade-off between DINO, CLIP-I, and CLIP-T across 29 subjects while requiring much less memory and lower rank than high rank LoRA versions. Code: https://github.com/donaldssh/NotAllLayersAreCreatedEqual.

Large Language Models (LLMs) struggle to keep up with the fast-changing nature of real-world information, as their pre-trained knowledge quickly becomes outdated. This work addresses the challenge of keeping LLMs up to date with factual knowledge (adaptation) while avoiding forgetting the relevant existing knowledge. Leveraging temporally-aligned Wikipedia and Wikidata dumps, we extract a continuous data stream and evaluate the performance of an incrementally trained GPT-2 across different time periods. Additionally, we extend our analysis to real-world news data using the RealTimeData dataset, examining how LLMs respond to novel facts, such as the COVID-19 pandemic. Our methodology includes synthetic data generation and SmartReview, a continual learning strategy that avoids forgetting by rehearsing on a carefully selected subset of the old data. Experimental results highlight that pretrained models require continual learning and demonstrate the effectiveness of replay-based approaches in mitigating forgetting. In particular, SmartReview provides a strong replay-based baseline that limits forgetting and enhances adaptation. This work advances the study of continual learning in LLMs, offering insights into the development of more temporally-aware and reliable AI systems.

In the wake of the COVID-19 pandemic, awareness of emerging pathogens has significantly increased, prompting greater investment in research and preparedness. In this context, arboviral diseases are recognized as unmet medical challenges due to their rapid spread. Notably, the geographical range of several flaviviral diseases is expanding: Zika virus (ZIKV), a member of the Flaviviridae family, has recently been linked to outbreaks associated with a rise in microcephaly cases in tropical regions. To contribute to the development of novel antiviral therapies, evaluation of a set of compounds with an antiviral activity against ZIKV was carried out. These compounds were originally identified as inhibitors of bovine viral diarrhea virus, another member of the Flaviviridae family. Two related compounds turned out to be active against ZIKV. One emerged as a particularly strong antiviral candidate, demonstrating high efficacy in inhibiting ZIKV replication, and became the focus of this study. Its activity was tested against a number of viruses of human health relevance and the compound was found to be effective against a number of viruses that use the endoplasmic reticulum as a replication hub. Indeed, we found that the Reticulon 3 protein is potently downregulated in the presence of the compound, whereas other endoplasmic reticulum-resident proteins are not affected. Because Reticulon 3 has a role in the replication of positive-sense single-stranded RNA viruses, an indirect antiviral effect of the compound studied was hypothesized. This compound may be considered as a promising lead for further studies aimed at the development of broad-spectrum antiviral drugs.

Self-supervised learning (SSL) is able to build latent representations that generalize well to unseen data. However, only a few SSL techniques exist for the online CL setting, where data arrives in small minibatches, the model must comply with a fixed computational budget, and task boundaries are absent. We introduce Continual Latent Alignment (CLA), a novel SSL strategy for Online CL that aligns the representations learned by the current model with past representations to mitigate forgetting. We found that our CLA is able to speed up the convergence of the training process in the online scenario, outperforming state-of-the-art approaches under the same computational budget. Surprisingly, we also discovered that using CLA as a pretraining protocol in the early stages of pretraining leads to a better final performance when compared to a full i.i.d. pretraining.

Pyridazin-3(2H)one-based molecules have always attracted the attention of medicinal chemists due to their different pharmacological properties. The incorporation of such nuclei in therapeutically active molecules either as monocyclic units or as fused bi- or tricyclic scaffolds results in a wide range of pharmacological effects such as anti-inflammatory, analgesic, anticancer, antimicrobial, antiviral, cardiovascular-protective, antiulcer, and many other useful pharmacological activities. In accordance with our consolidated experience gained over the years in the chemistry and biology of tricyclic pyridazin-3(2H)ones, this review summarizes SAR studies of such pyridazinone-based polycyclic compounds endowed with various biological and therapeutic properties.

Online Continual Learning (OCL) methods train a model on a non-stationary data stream where only a few examples are available at a time, often leveraging replay strategies. However, usage of replay is sometimes forbidden, especially in applications with strict privacy regulations. Therefore, we propose Continual MultiPatches (CMP), an effective plug-in for existing OCL self-supervised learning strategies that avoids the use of replay samples. CMP generates multiple patches from a single example and projects them into a shared feature space, where patches coming from the same example are pushed together without collapsing into a single point. CMP surpasses replay and other SSL-based strategies on OCL streams, challenging the role of replay as a go-to solution for self-supervised OCL.

A series of N-acylhydrazones bearing a 1,4-dihydroindeno[1,2-b]pyrrole ring, along with benzene and thiophene rings substituted with chlorine or methyl groups, was synthesized and evaluated for their antiproliferative and cytotoxic activity against the melanoma A375 cell line and to measure the inhibition of tubulin polymerization. Four compounds elicited interesting activity: derivatives, 1g and 1h showed a 25% slowdown of tubulin polymerization, whereas compounds 2c and 2d caused a slowdown of 40% and 60%, respectively. Molecular modelling results have confirmed that the most active N-acylhydrazones (1g, 1h, 2c, and 2d) may act as tubulin polymerization inhibitors.

In this work, we synthesized a small library of tricyclic compounds to assess whether they might have both anticancer and antiviral activity against three viruses that have recently caused epidemics. Concerning their anti-tumour activity, derivative 1 was found to be the compound with the highest GI50 values on some cancer cell line panels. Particularly, in melanoma cell lines, its GI50 values ranged between 1.54 μM (MALME-3M) and 2.03 μM (M14). Several derivatives with considerable anti-tumour activity showed antiviral activity as well, against influenza A virus (e.g., derivative 19, selectivity index of 21.36 in MDCK cells) or against Zika virus (compound 13, selectivity index of 20.20 in Huh-7 cells). Moreover, compounds 13 and 12 showed anti-SARS-CoV-2 activity, with selectivity indices of 150.00 and 63.63, respectively. Compound 1, for its anticancer activity, and 13, for its anti-SARS-CoV-2 activity, together with the compounds active against Zika virus and influenza A virus, are promising candidates for further studies.

The contemporary AI landscape demands a holistic framework to ensure security across the entire AI supply chain and lifecycle. Despite the availability of existing adversarial attack techniques, an end-to-end solution for identifying threats, vulnerabilities, and risks is still lacking. Despite EU initiatives like the AI Act promoting safety and trustworthiness in AI, it lacks a system for managing weaknesses within a networked AI supply chain. This paper introduces CoEvolution, which aspires to address this gap by implementing a complete Security, Trust, and Robustness (STR) assessment solution, capable of addressing evolving AI cybersecurity threats. CoEvolution proposes a universal hub for STR risk assessment and security assurance, aligned with MLDevOps practices and EU AI regulatory frameworks. It introduces innovative AI model descriptions, including an AI Model Bill of Materials, coupled with security monitoring and context awareness. CoEvolution seeks to ensure compliance with EU directives on trust, fairness, data governance, and GDPR guidelines.

Continual learning (CL) is the sub-field of machine learning concerned with accumulating knowledge in dynamic environments. So far, CL research has mainly focused on incremental classification tasks, where models learn to classify new categories while retaining knowledge of previously learned ones. Here, we argue that maintaining such a focus limits both theoretical development and practical applicability of CL methods. Through a detailed analysis of concrete examples - including multi-target classification, robotics with constrained output spaces, learning in continuous task domains, and higher-level concept memorization - we demonstrate how current CL approaches often fail when applied beyond standard classification. We identify three fundamental challenges: (C1) the nature of continuity in learning problems, (C2) the choice of appropriate spaces and metrics for measuring similarity, and (C3) the role of learning objectives beyond classification. For each challenge, we provide specific recommendations to help move the field forward, including formalizing temporal dynamics through distribution processes, developing principled approaches for continuous task spaces, and incorporating density estimation and generative objectives. In so doing, this position paper aims to broaden the scope of CL research while strengthening its theoretical foundations, making it more applicable to real-world problems.

Online Continual Learning (OCL) models continuously adapt to nonstationary data streams, usually without task information. These settings are complex and many traditional CL methods fail, while online methods (mainly replay-based) suffer from instabilities after the task shift. To address this issue, we formalize replay-based OCL as a second-order online joint optimization with explicit KL-divergence constraints on replay data. We propose Online Curvature-Aware Replay (OCAR) to solve the problem: a method that leverages second-order information of the loss using a K-FAC approximation of the Fisher Information Matrix (FIM) to precondition the gradient. The FIM acts as a stabilizer to prevent forgetting while also accelerating the optimization in non-interfering directions. We show how to adapt the estimation of the FIM to a continual setting, stabilizing second-order optimization for non-iid data, uncovering the role of the Tikhonov damping in the stability-plasticity tradeoff. Empirical results show that OCAR outperforms state-of-the-art methods in continual metrics, achieving higher average accuracy throughout the training process in three different benchmarks.

A small library of 25 novel 1,3,4-oxadiazole-quinoxalines was synthesized and evaluated in vitro for its cytotoxic activity at 10 μM concentration against nine NCI-different cancer cell lines. Among tested compounds, derivatives 24, 25, and 26 showed good inhibition percentages over different cell lines and, therefore, progressed to the full five-dose assay. Compound 24, possessing a 1,3,4-oxadiazole-core, bearing a 7-trifluoromethyl-quinoxaline nucleus on C-2 and a C-5 phenyl ring, had activity against leukemia, CNS, ovarian, renal, prostate, and breast cancer, with highest the values against breast MCF7 (GI50: 1.85 μM) and MDA-MB-468 (GI50: 1.95 μM) cell lines, showing the better MG_MID value (-5.02). These novel derivatives were able to delay the S phase of the cell cycle and induce apoptosis.