Abstract
This work introduces Gen-JEMA, a generative approach based on joint embedding with multimodal alignment (JEMA), to enhance feature extraction in the embedding space and improve the explainability of its predictions. Gen-JEMA addresses these challenges by leveraging multimodal data, including multi-view images and metadata such as process parameters, to learn transferable semantic representations. Gen-JEMA enables more explainable and enriched predictions by learning a decoder from the embedding. This novel co-learning framework, tailored for directed energy deposition (DED), integrates multiple data sources to learn a unified data representation and predict melt pool images from the primary sensor. The proposed approach enables real-time process monitoring using only the primary modality, simplifying hardware requirements and reducing computational overhead. The effectiveness of Gen-JEMA for DED process monitoring was evaluated, focusing on its generalization to downstream tasks such as melt pool geometry prediction and the generation of external melt pool representations using off-axis sensor data. To generate these external representations, autoencoder (AE) and variational autoencoder (VAE) architectures were optimized using Bayesian optimization. The AE outperformed other approaches achieving a 38% improvement in melt pool geometry prediction compared to the baseline and 88% in data generation compared with the VAE. The proposed framework establishes the foundation for integrating multisensor data with metadata through a generative approach, enabling various downstream tasks within the DED domain and achieving a small embedding, allowing efficient process control based on model predictions and embeddings. © The Author(s) 2025.

Abstract
Text simplification converts complex text into simpler language, improving readability and comprehension. This study evaluates the effectiveness of open-source large language models for text simplification across various categories. We created a dataset of 66,620 lead section pairs from English and Simple English Wikipedia, spanning nine categories, and tested Llama 3 for text simplification. We assessed its output for readability, simplicity, and meaning preservation. Results show improved readability, with simplification varying by category. Texts on Time were the most shortened, while Leisurerelated texts had the greatest reduction of words/characters and syllables per sentence. Meaning preservation was most effective for the Objects and Education categories.

Abstract
The Abstraction and Reasoning Corpus (ARC-AGI) is an Artificial General Intelligence benchmark that is currently unsolved. It demands strong generalization and reasoning capabilities, which are known to be weaknesses of Neural Network based systems. In this work, we propose a Program synthesis system to solve it, which casts an ARC-AGI task as a sequence of Inductive Logic Programming tasks. We have implemented a simple Domain Specific Language that corresponds to a small set of object-centric abstractions relevant to the benchmark. This allows for adequate representations to be used to create logic programs, which provide reasoning capabilities to our system. When solving each task, the proposed system can generalize from a few training pairs of input-output grids. The obtained logic programs are able to generate objects present in the output grids and can transform the test input grid into the output grid solution. We developed our system based on some ARC-AGI tasks that do not require more than the small number of primitives that we implemented and showed that our system can solve unseen tasks that require different reasoning.

Abstract
Entity linking is an important task in medical natural language processing (NLP) for converting unstructured text into structured data for clinical analysis and semantic interoperability. However, in lower-resource languages, this task is challenging due to the limited availability of domain-specific resources. This paper explores a translation-based cross-lingual entity linking approach using GPT models, GPT-3.5 and GPT-4o, for zero-shot machine translation and entity linking with in-context learning. We evaluate our approach using a Portuguese-English parallel dataset of radiology abstracts. Our results show that chunk-level machine translation outperforms sentence-level translation. Moreover, our translation-based approach to cross-lingual entity linking of UMLS concepts outperformed the multilingual encoder method baseline. However, the in-context learning entity linking approach did not outperform a translation-based approach with a dictionary-based entity linking method. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Abstract
Driven by the NIST's post-quantum standardization efforts and the selection of Kyber as a lattice-based Key-Encapsulation Mechanism (KEM), severalPasswordAuthenticated KeyExchange (PAKE) protocols have been recently proposed that leverage a KEM to create an efficient, easy-to-implement and secure PAKE. In two recent works, Beguinet et al. (ACNS 2023) and Pan and Zeng (ASIACRYPT 2023) proposed generic compilers that transform KEM into PAKE, relying on an Ideal Cipher (IC) defined over a group. However, although IC on a group is often used in cryptographic protocols, special care must be taken to instantiate such objects in practice, especially when a low-entropy key is used. To address this concern, Dos Santos et al. (EUROCRYPT 2023) proposed a relaxation of the ICmodel under the Universal Composability (UC) framework called Half-Ideal Cipher (HIC). They demonstrate how to construct a UC-secure PAKE protocol, EKE-KEM, from a KEM and a modified 2round Feistel construction called m2F. Remarkably, the m2F sidesteps the use of an IC over a group, and instead employs an IC defined over a fixed-length bitstring domain, which is easier to instantiate. In this paper, we introduce a novel PAKE protocol called CHIC that improves the communication and computation efficiency of EKE-KEM, by avoiding the HIC abstraction. Instead, we split the KEM public key in two parts and use the m2F directly, without further randomization. We provide a detailed proof of the security of CHIC and establish precise security requirements for the underlying KEM, including one-wayness and anonymity of ciphertexts, and uniformity of public keys. Our findings extend to general KEM-based EKE-style protocols and show that a passively secure KEM is not sufficient. In this respect, our results align with those of Pan and Zeng (ASIACRYPT 2023), but contradict the analyses of KEM-to-PAKE compilers by Beguinet et al. (ACNS 2023) and Dos Santos et al. (EUROCRYPT 2023). Finally, we provide an implementation of CHIC, highlighting its minimal overhead compared to the underlying KEM - Kyber. An interesting aspect of the implementation is that we reuse the rejection sampling procedure in Kyber reference code to address the challenge of hashing onto the public key space. As of now, to the best of our knowledge, CHIC stands as the most efficient PAKE protocol from black-box KEM that offers rigorously proven UC security.

Abstract
Here we define a new unification algorithm for terms interpreted in semantic domains denoted by a subclass of regular types here called deterministic regular types. This reflects our intention not to handle the semantic universe as a homogeneous collection of values, but instead, to partition it in a way that is similar to data types in programming languages. We first define the new unification algorithm which is based on constraint generation and constraint solving, and then prove its main properties: termination, soundness, and completeness with respect to the semantics. Finally, we discuss how to apply this algorithm to a dynamically typed version of Prolog.