Invasive mechanical ventilation (MV) is a life-sustaining therapy commonly used in the intensive care unit (ICU) for patients with severe and acute conditions. These patients frequently rely on MV for breathing. Given the high risk of death in such cases, optimal MV settings can reduce mortality, minimize ventilator-induced lung injury, shorten ICU stays, and ease the strain on healthcare resources. However, optimizing MV settings remains a complex and error-prone process due to patient-specific variability. While Offline Reinforcement Learning (RL) shows promise for optimizing MV settings, current methods struggle with the hybrid (continuous and discrete) nature of MV settings. Discretizing continuous settings leads to exponential growth in the action space, which limits the number of optimizable settings. Converting the predictions back to continuous can cause a distribution shift, compromising safety and performance. To address this challenge, in the IntelliLung project, we are developing an AI-based approach where we constrain the action space and employ factored action critics. This approach allows us to scale to six optimizable settings compared to 2-3 in previous studies. We adapt SOTA offline RL algorithms to operate directly on hybrid action spaces, avoiding the pitfalls of discretization. We also introduce a clinically grounded reward function based on ventilator-free days and physiological targets. Using multi-objective optimization for reward selection, we show that this leads to a more equitable consideration of all clinically relevant objectives. Notably, we develop a system in close collaboration with healthcare professionals that is aligned with real-world clinical objectives and designed with future deployment in mind.

International audience

Abstract Generating context-aware, high-quality, and innovative scientific ideas remains a central challenge in AI-supported research. We introduce SCI-IDEA , a two-stage framework combining large language models (LLMs) with a specialised Aha-Moment Detection module for iterative idea refinement. The first stage extracts structured facets: objectives, methodology, evaluation, and future work from previous publications, compressing each paper to $${\sim }$$ 200 tokens to enable scalable processing of researcher profiles that would otherwise exceed frontier LLM context windows. The second stage integrates these facets with a token-level embedding approach to identify novelty and context gaps, indicating candidate ideas as Aha moments when they surpass predefined thresholds for both novelty and surprise. We evaluate SCI-IDEA across 100 computer-science researcher profiles, 4 LLMs (GPT-4o, GPT−4.5, DeepSeek-32B, DeepSeek-70B), three embedding strategies, and 5 prompting configurations via a hybrid protocol of automated LLM-as-judge scoring with GPT−4.1 and a human evaluation with 15 PhD-level domain experts. These two evaluation signals diverge substantially: LLM-based scores exceed expert ratings by 3–4 points on a 10-point scale, with near-zero inter-rater correlation ( $$r = 0.02$$ –0.17, $$p > 0.05$$ ), indicating that automated scores reflect relative architectural comparisons rather than human-verified measures of absolute idea quality. Within the LLM-evaluated setting, SCI-IDEA with token-level embeddings achieves a mean quality score of 6.97, a modest but constant improvement over the LLM-only baseline ( $$\Delta = +0.20$$ points; $$p < 0.05$$ ) and driven primarily by improved feasibility scores ( $$+0.21$$ points, $$p < 0.001$$ ). Expert evaluators independently confirmed this trend while assigning lower absolute scores. These findings suggest an architectural advantage of facet-based context modelling and token-level novelty detection, though their practical importance deserves validation through longitudinal studies with larger domain experts. This work is validated in the computer science domain, but its application to other fields requires further investigation.

Knowledge graph embedding (KGE) models provide a low-dimensional representation of knowledge graphs in continuous vector spaces. This representation learning enables different downstream AI tasks such as link prediction for graph completion. However, most embedding models are only designed considering the algebra and geometry of the entity embedding space, the algebra of the relation embedding space, and the interaction between relation and entity embeddings. Neglecting the geometry of relation embedding limits the optimization of entity and relation distribution leading to suboptimal performance of knowledge graph completion. To address this issue, we propose a new perspective in the design of KGEs by looking into the geometry of relation embedding space. The proposed method and its variants are developed on top of an existing framework, RotatE, from which we leverage the geometry of the relation embeddings by mutating the unit circle to an ellipse, and further generalize it with the concept of a butterfly curve, consecutively. Besides the theoretical abilities of the model in preserving topological and relational patterns, the experiments on the WN18RR, FB15K-237 and YouTube benchmarks showed that this new family of KGEs can challenge or outperform state-of-the-art models.

This study introduces a Model-Based Deep Reinforcement Learning approach to enhance the effectiveness and transparency of mechanical ventilation treatment in the critical care setting of Intensive Care Units (ICUs). Distinct from conventional model-free methods, our approach benefits from the model-based algorithms’ capability to learn and interrogate dynamics models, enabling better generalization through synthetic data generation and a deeper understanding of the system dynamics. Coupled with Explainable AI (XAI) techniques, we focus on uncovering the underlying mechanisms of patient-ventilator interactions as learned by the AI. Our findings show a significant improvement in treatment efficacy, measured by Fitted Q Evaluation (FQE) metrics, achieved without the need for auxiliary rewards. This advancement not only highlights the potential of model-based reinforcement learning in healthcare but also emphasizes the importance of transparent AI design in healthcare applications.

Abstract Answering factual questions from heterogenous sources, such as graphs and text, is a key capacity of intelligent systems. Current approaches either (i) perform question answering over text and structured sources as separate pipelines followed by a merge step or (ii) provide an early integration, giving up the strengths of particular information sources. To solve this problem, we present “HumanIQ”, a method that teaches language models to dynamically combine retrieved information by imitating how humans use retrieval tools. Our approach couples a generic method for gathering human demonstrations of tool use with adaptive few-shot learning for tool augmented models. We show that HumanIQ confers significant benefits, including i) reducing the error rate of our strongest baseline (GPT-4) by over 50% across 3 benchmarks, (ii) improving human preference over responses from vanilla GPT-4 (45.3% wins, 46.7% ties, 8.0% loss), and (iii) outperforming numerous task-specific baselines.

Skilled employees are the most important pillars of an organization. Despite this, most organizations face high attrition and turnover rates. While several machine learning models have been developed to analyze attrition and its causal factors, the interpretations of those models remain opaque. In this paper, we propose the HR-DSS approach, which stands for Human Resource (HR) Decision Support System, and uses explainable AI for employee attrition problems. The system is designed to assist HR departments in interpreting the predictions provided by machine learning models. In our experiments, we employ eight machine learning models to provide predictions. We further process the results achieved by the best-performing model by the SHAP explainability process and use the SHAP values to generate natural language explanations which can be valuable for HR. Furthermore, using "What-if-analysis", we aim to observe plausible causes for attrition of an individual employee. The results show that by adjusting the specific dominant features of each individual, employee attrition can turn into employee retention through informative business decisions.

We propose the use of controlled natural language as a target for knowledge graph question answering (KGQA) semantic parsing via language models as opposed to using formal query languages directly. Controlled natural languages are close to (human) natural languages, but can be unambiguously translated into a formal language such as SPARQL. Our research hypothesis is that the pre-training of large language models (LLMs) on vast amounts of textual data leads to the ability to parse into controlled natural language for KGQA with limited training data requirements. We devise an LLM-specific approach for semantic parsing to study this hypothesis. To conduct our study, we created a dataset that allows the comparison of one formal and two different controlled natural languages. Our analysis shows that training data requirements are indeed substantially reduced when using controlled natural languages, which is relevant since collecting and maintaining high-quality KGQA semantic parsing training data is very expensive and time-consuming.

Subspace distance is an invaluable tool exploited in a wide range of feature selection methods. The power of subspace distance is that it can identify a representative subspace, including a group of features that can efficiently approximate the space of original features. On the other hand, employing intrinsic statistical information of data can play a significant role in a feature selection process. Nevertheless, most of the existing feature selection methods founded on the subspace distance are limited in properly fulfilling this objective. To pursue this void, we propose a framework that takes a subspace distance into account which is called "Variance-Covariance subspace distance". The approach gains advantages from the correlation of information included in the features of data, thus determines all the feature subsets whose corresponding Variance-Covariance matrix has the minimum norm property. Consequently, a novel, yet efficient unsupervised feature selection framework is introduced based on the Variance-Covariance distance to handle both the dimensionality reduction and subspace learning tasks. The proposed framework has the ability to exclude those features that have the least variance from the original feature set. Moreover, an efficient update algorithm is provided along with its associated convergence analysis to solve the optimization side of the proposed approach. An extensive number of experiments on nine benchmark datasets are also conducted to assess the performance of our method from which the results demonstrate its superiority over a variety of state-of-the-art unsupervised feature selection methods. The source code is available at https://github.com/SaeedKarami/VCSDFS.

Declarative rules such as Prolog and Datalog rules are common formalisms to express expert knowledge and facts. They play an important role in Knowledge Graph (KG) construction and completion. Such rules not only encode the expert background knowledge and the relational patterns among the data, but also infer new knowledge and insights from them. Formalizing rules is often a laborious manual process, while learning them from data automatically can ease this process. Within the rule hypothesis space, current approaches resort to exhaustive search with a number of heuristics and syntactic restrictions on the rule language, which impacts the efficiency and quality of the outcome rules. In this paper, we extend the rule hypothesis space from usual path rules to general Datalog rule space by proposing a novel Genetic Logic Programming algorithm named Evoda. It is an iterative process to learn high-quality rules over large scale KG for a matter of seconds. We have performed experiments over multiple real-world KGs and various evaluation metrics to show its mining capabilities for higher quality rules and more precise predictions. Additionally, we have applied it on the KG completion tasks to illustrate its competitiveness with several state-of-the-art embedding or neural-based models. The experiments demonstrate the feasibility, effectiveness and efficiency of the Evoda algorithm.

9064

Knowledge Graph Embeddings (KGEs) have shown promising performance on link prediction tasks by mapping the entities and relations from a knowledge graph into a geometric space.The capability of KGEs in preserving graph characteristics including structural aspects and semantics, highly depends on the design of their score function, as well as the inherited abilities from the underlying geometry.Many KGEs use the Euclidean geometry which renders them incapable of preserving complex structures and consequently causes wrong inferences by the models.To address this problem, we propose a neuro differential KGE that embeds nodes of a KG on the trajectories of Ordinary Differential Equations (ODEs).To this end, we represent each relation (edge) in a KG as a vector field on several manifolds.We specifically parameterize ODEs by a neural network to represent complex manifolds and complex vector fields on the manifolds.Therefore, the underlying embedding space is capable to assume the shape of various geometric forms to encode heterogeneous subgraphs.Experiments on synthetic and benchmark datasets using state-of-the-art KGE models justify the ODE trajectories as a means to enable structure preservation and consequently avoiding wrong inferences.

Knowledge graphs (KGs) represent world facts in a structured form. Although knowledge graphs are quantitatively huge and consist of millions of triples, the coverage is still only a small fraction of world's knowledge. Among the top approaches of recent years, link prediction using knowledge graph embedding (KGE) models has gained significant attention for knowledge graph completion. Various embedding models have been proposed so far, among which, some recent KGE models obtain state-of-the-art performance on link prediction tasks by using embeddings with a high dimension (e.g. 1000) which accelerate the costs of training and evaluation considering the large scale of KGs. In this paper, we propose a simple but effective performance boosting strategy for KGE models by using multiple low dimensions in different repetition rounds of the same model. For example, instead of training a model one time with a large embedding size of 1200, we repeat the training of the model 6 times in parallel with an embedding size of 200 and then combine the 6 separate models for testing while the overall numbers of adjustable parameters are same (6*200=1200) and the total memory footprint remains the same. We show that our approach enables different models to better cope with their expressiveness issues on modeling various graph patterns such as symmetric, 1-n, n-1 and n-n. In order to justify our findings, we conduct experiments on various KGE models. Experimental results on standard benchmark datasets, namely FB15K, FB15K-237 and WN18RR, show that multiple low-dimensional models of the same kind outperform the corresponding single high-dimensional models on link prediction in a certain range and have advantages in training efficiency by using parallel training while the overall numbers of adjustable parameters are same.

Improvements in web technologies and artificial intelligence enable novel, more data-driven research practices for scientists. However, scientific knowledge generated from data-intensive research practices is disseminated with unstructured formats, thus hindering the scholarly communication in various respects. The traditional document-based representation of scholarly information hampers the reusability of research contributions. To address this concern, we developed the Physics Ontology (PhySci) to represent physics-related scholarly data in a machine-interpretable format. PhySci facilitates knowledge exploration, comparison, and organization of such data by representing it as knowledge graphs. It establishes a unique conceptualization to increase the visibility and accessibility to the digital content of physics publications. We present the iterative design principles by outlining a methodology for its development and applying three different evaluation approaches: data-driven and criteria-based evaluation, as well as ontology testing.

Embedding models based on translation and rotation have gained significant attention in link prediction tasks for knowledge graphs. Most of the earlier works have modified the score function of Knowledge Graph Embedding models in order to improve the performance of link prediction tasks. However, as proven theoretically and experimentally, the performance of such Embedding models strongly depends on the loss function. One of the prominent approaches in defining loss functions is to set a margin between positive and negative samples during the learning process. This task is particularly important because it directly affects the learning and ranking of triples and ultimately defines the final output. Approaches for setting a margin have the following challenges: a) the length of the margin has to be fixed manually, b) without a fixed point for center of the margin, the scores of positive triples are not necessarily enforced to be sufficiently small to fulfill the translation/rotation from head to tail by using the relation vector. In this paper, we propose a family of loss functions dubbed SlidE±to address the aforementioned challenges. The formulation of the proposed loss functions enables an automated technique to adjust the length of the margin adaptive to a defined center. In our experiments on a set of standard benchmark datasets including Freebase and WordNet, the effectiveness of our approach is confirmed for training Knowledge Graph Embedding models, specifically TransE and RotatE as a case study, on link prediction tasks.

A vast amount of OpenCourseWare (OCW) is meanwhile being published online to make educational content accessible to larger audiences. The awareness of such courses among users and the popularity of systems providing such courses are increasing. However, from a subjective experience, OCW is frequently cursory, outdated or non-reusable. In order to obtain a better understanding of the quality of OCW, we assess the quality in terms of fitness for use. Based on three OCW use case scenarios, we define a range of dimensions according to which the quality of courses can be measured. From the definition of each dimension a comprehensive list of quality metrics is derived. In order to obtain a representative overview of the quality of OCW, we performed a quality assessment on a set of 100 randomly selected courses obtained from 20 different OCW repositories. Based on this assessment we identify crucial areas in which OCW needs to improve in order to deliver up to its promises.