Is reading always better for your brain than listening to audiobooks? Reading books and listening to audiobooks tap into different elements of cognition, each with their own benefits. So which one should you choose, and when? But when a friend recently asked me whether her daughter was getting the same cognitive benefits from an audiobook as she would from reading, my instinct was to think "she's enjoying a book, the format doesn't matter". However, when I dug into the science, I found the medium does shape the mind in subtly different but meaningful ways.

'Let's talk about something fun!' Grem the AI chatbot toy. 'Let's talk about something fun!' Grem the AI chatbot toy. 'I love you too!' My family's creepy, unsettling week with an AI toy The cuddly chatbot Grem is designed to'learn' your child's personality, while every conversation they have is recorded, then transcribed by a third party. It wasn't long before I wanted this experiment to be over ... 'I'm going to throw that thing into a river!" my wife says as she comes down the stairs looking frazzled after putting our four-year-old daughter to bed. To be clear, "that thing" is not our daughter, Emma*. It's Grem, an AI-powered stuffed alien toy that the musician Claire Boucher, better known as Grimes, helped develop with toy company Curio. Designed for kids aged three and over and built with OpenAI's technology, the toy is supposed to "learn" your child's personality and have fun, educational conversations with them. It's advertised as a healthier alternative to screen time and is ...

Data factorization methods have met with considerable success in discovering latent features of the signals encountered in wide-ranging applications. In this way, the representation bases, which make up the columns of the basis matrix or dictionary, are learned from the available samples of the target environment. An example is the sparse representation (SR) in which the dictionary is intended to best represent the data with a small number of atoms, much smaller than the dimension of the signal space. It has been shown that, in addition to a more informative representation of signals, imposing sparsity constraints on the representation coefficients can improve the generalization performance and the computational efficiency [1, 2, 3]. Furthermore, the sparse representation is more robust to noise, redundancy, and missing data. These features are mainly attributed to the fact that the intrinsic dimension of natural signals is usually much smaller than their apparent dimension and hence SR in an appropriate dictionary can extract these intrinsic features more efficiently. SR has been a successful strategy and has received considerable attention and achieved state-of-the-art results in many applications, e.g.

Language models (LMs) have demonstrated remarkable capabilities in NLP, yet adapting them efficiently and robustly to specific tasks remains challenging. As their scale and complexity grow, fine-tuning LMs on labelled data often underutilizes available unlabelled data, leads to overfitting on small task-specific sets, and imposes significant computational costs. These limitations hamper their application to the open-ended landscape of real-world language tasks. This thesis proposes a series of methods to better adapt LMs to downstream applications. First, we explore strategies for extracting task-relevant knowledge from unlabelled data, introducing a novel continued pre-training technique that outperforms state-of-the-art semi-supervised approaches. Next, we present a parameter-efficient fine-tuning method that substantially reduces memory and compute costs while maintaining competitive performance. We also introduce improved supervised fine-tuning methods that enable LMs to better follow instructions, especially when labelled data is scarce, enhancing their performance across a range of NLP tasks, including open-ended generation. Finally, we develop new evaluation methods and benchmarks, such as multi-hop spatial reasoning tasks, to assess LM capabilities and adaptation more comprehensively. Through extensive empirical studies across diverse NLP tasks, our results demonstrate that these approaches substantially improve LM robustness, efficiency, and generalization, making them more adaptable to a broad range of applications. These advances mark a significant step towards more robust and efficient LMs, bringing us closer to the goal of artificial general intelligence.

Large language models (LLMs) offer unprecedented and growing capabilities, but also introduce complex safety and security challenges that resist conventional risk management. While conventional probabilistic risk analysis (PRA) requires exhaustive risk enu meration and quantification, the novelty and complexity of these systems make PRA impractical, particularly against adaptive adversaries. Previous research found that risk management in various fields of engineering such as nuclear or civil engineering is often solved by generic (i.e. Here we show how emerging risks in LLM - powered systems could be met with 100+ of these non - probabilistic strategies to risk management, including risks from adaptive adversaries. The strategies are divided into five categories and are mapped to LLM secur ity (and AI safety more broadly). We also present an LLM - powered workflow for applying these strategies and other workflows suitable for solution architec ts. Overall, these strategies could contribute (despite some limitations) to security, safety and other dimensions of responsible AI.

Traditional gradient-based sampling methods, like standard Hamiltonian Monte Carlo, require that the desired target distribution is continuous and differentiable. This limits the types of models one can define, although the presented models capture the reality in the observations better. In this project, Generalized Randomized Hamiltonian Monte Carlo (GRHMC) processes for sampling continuous densities with discontinuous gradient and piecewise smooth targets are proposed. The methods combine the advantages of Hamiltonian Monte Carlo methods with the nature of continuous time processes in the form of piecewise deterministic Markov processes to sample from such distributions. It is argued that the techniques lead to GRHMC processes that admit the desired target distribution as the invariant distribution in both scenarios. Simulation experiments verifying this fact and several relevant real-life models are presented, including a new parameterization of the spike and slab prior for regularized linear regression that returns sparse coefficient estimates and a regime switching volatility model.

Recent advancements in large language models (LLMs) have enabled their successful application to a broad range of tasks. However, in information-intensive tasks, the prompt length can grow fast, leading to increased computational requirements, performance degradation, and induced biases from irrelevant or redundant information. Recently, various prompt compression techniques have been introduced to optimize the trade-off between reducing input length and retaining performance. We propose a holistic evaluation framework that allows for in-depth analysis of prompt compression methods. We focus on three key aspects, besides compression ratio: (i) downstream task performance, (ii) grounding in the input context, and (iii) information preservation. Through this framework, we investigate state-of-the-art soft and hard compression methods, showing that they struggle to preserve key details from the original prompt, limiting their performance on complex tasks. We demonstrate that modifying soft prompting methods to control better the granularity of the compressed information can significantly improve their effectiveness -- up to +23\% in downstream task performance, more than +8 BERTScore points in grounding, and 2.7x more entities preserved in compression.

Retrieval-augmented generation (RAG) faces challenges related to factual correctness, source attribution, and response completeness. To address them, we propose a modular pipeline for grounded response generation that operates on information nuggets-minimal, atomic units of relevant information extracted from retrieved documents. The multistage pipeline encompasses nugget detection, clustering, ranking, top cluster summarization, and fluency enhancement. It guarantees grounding in specific facts, facilitates source attribution, and ensures maximum information inclusion within length constraints. Extensive experiments on the TREC RAG'24 dataset evaluated with the AutoNuggetizer framework demonstrate that GINGER achieves state-of-the-art performance on this benchmark.

Federated Learning (FL) enables privacy-preserving multi-source information fusion (MSIF) but is challenged by client drift in highly heterogeneous data settings. Many existing drift-mitigation strategies rely on reference-based techniques--such as gradient adjustments or proximal loss--that use historical snapshots (e.g., past gradients or previous global models) as reference points. When only a subset of clients participates in each training round, these historical references may not accurately capture the overall data distribution, leading to unstable training. In contrast, our proposed Gradient Centralized Federated Learning (GC-Fed) employs a hyperplane as a historically independent reference point to guide local training and enhance inter-client alignment. GC-Fed comprises two complementary components: Local GC, which centralizes gradients during local training, and Global GC, which centralizes updates during server aggregation. In our hybrid design, Local GC is applied to feature-extraction layers to harmonize client contributions, while Global GC refines classifier layers to stabilize round-wise performance. Theoretical analysis and extensive experiments on benchmark FL tasks demonstrate that GC-Fed effectively mitigates client drift and achieves up to a 20% improvement in accuracy under heterogeneous and partial participation conditions.

The CheckThat! lab aims to advance the development of innovative technologies designed to identify and counteract online disinformation and manipulation efforts across various languages and platforms. The first five editions focused on key tasks in the information verification pipeline, including check-worthiness, evidence retrieval and pairing, and verification. Since the 2023 edition, the lab has expanded its scope to address auxiliary tasks that support research and decision-making in verification. In the 2025 edition, the lab revisits core verification tasks while also considering auxiliary challenges. Task 1 focuses on the identification of subjectivity (a follow-up from CheckThat! 2024), Task 2 addresses claim normalization, Task 3 targets fact-checking numerical claims, and Task 4 explores scientific web discourse processing. These tasks present challenging classification and retrieval problems at both the document and span levels, including multilingual settings.