Limited infrastructure, scarce educational resources, and unreliable internet access often hinder physics and photonics education in underdeveloped regions. These barriers create deep inequities in Science, Technology, Engineering, and Mathematics (STEM) education. This article explores how Small Language Models (SLMs)-compact, AI-powered tools that can run offline on low-power devices, offering a scalable solution. By acting as virtual tutors, enabling native-language instruction, and supporting interactive learning, SLMs can help address the shortage of trained educators and laboratory access. By narrowing the digital divide through targeted investment in AI technologies, SLMs present a scalable and inclusive solution to advance STEM education and foster scientific empowerment in marginalized communities.

--The analysis of conversational dynamics has gained increasing importance with the rise of large language model-based systems, which interact with users across diverse contexts. In this work, we propose a novel computational framework for constructing conversational graphs that capture the flow and structure of loosely organized dialogues, referred to as quasi-patterned conversations. We introduce the Filter & Reconnect method, a novel graph simplification technique that minimizes noise while preserving semantic coherence and structural integrity of conversational graphs. Through comparative analysis, we demonstrate that the use of large language models combined with our graph simplification technique has resulted in semantic metric S increasing by a factor of 2.06 compared to previous approaches while simultaneously enforcing a tree-like structure with 0 δ -hyperbolicity, ensuring optimal clarity in conversation modeling. This work provides a computational method for analyzing large-scale dialogue datasets, with practical applications related to monitoring automated systems such as chatbots, dialogue management tools, and user behavior analytics.

Developing narrative and comprehension skills in early childhood is critical for later literacy. However, teachers in large preschool classrooms struggle to accurately identify students who require intervention. We present a system for automatically assessing oral narratives of preschool children in Afrikaans and isiXhosa. The system uses automatic speech recognition followed by a machine learning scoring model to predict narrative and comprehension scores. For scoring predicted transcripts, we compare a linear model to a large language model (LLM). The LLM-based system outperforms the linear model in most cases, but the linear system is competitive despite its simplicity. The LLM-based system is comparable to a human expert in flagging children who require intervention. We lay the foundation for automatic oral assessments in classrooms, giving teachers extra capacity to focus on personalised support for children's learning.

Oral narrative skills are strong predictors of later literacy development. This study examines the features of oral narratives from children who were identified by experts as requiring intervention. Using simple machine learning methods, we analyse recorded stories from four- and five-year-old Afrikaans- and isiXhosa-speaking children. Consistent with prior research, we identify lexical diversity (unique words) and length-based features (mean utterance length) as indicators of typical development, but features like articulation rate prove less informative. Despite cross-linguistic variation in part-of-speech patterns, the use of specific verbs and auxiliaries associated with goal-directed storytelling is correlated with a reduced likelihood of requiring intervention. Our analysis of two linguistically distinct languages reveals both language-specific and shared predictors of narrative proficiency, with implications for early assessment in multilingual contexts.

In observational studies, confounding variables affect both treatment and outcome. Moreover, instrumental variables also influence the treatment assignment mechanism. This situation sets the study apart from a standard randomized controlled trial, where the treatment assignment is random. Due to this situation, the estimated average treatment effect becomes biased. To address this issue, a standard approach is to incorporate the estimated propensity score when estimating the average treatment effect. However, these methods incur the risk of misspecification in propensity score models. To solve this issue, a novel method called the "Self balancing neural network" (Sbnet), which lets the model itself obtain its pseudo propensity score from the balancing net, is proposed in this study. The proposed method estimates the average treatment effect by using the balancing net as a key part of the feedforward neural network. This formulation resolves the estimation of the average treatment effect in one step. Moreover, the multi-pseudo propensity score framework, which is estimated from the diversified balancing net and used for the estimation of the average treatment effect, is presented. Finally, the proposed methods are compared with state-of-the-art methods on three simulation setups and real-world datasets. It has been shown that the proposed self-balancing neural network shows better performance than state-of-the-art methods.

The Sliced Gromov-Wasserstein (SGW) distance, aiming to relieve the computational cost of solving a non-convex quadratic program that is the Gromov-Wasserstein distance, utilizes projecting directions sampled uniformly from unit hyperspheres. This slicing mechanism incurs unnecessary computational costs due to uninformative directions, which also affects the representative power of the distance. However, finding a more appropriate distribution over the projecting directions (slicing distribution) is often an optimization problem in itself that comes with its own computational cost. In addition, with more intricate distributions, the sampling itself may be expensive. As a remedy, we propose an optimization-free slicing distribution that provides fast sampling for the Monte Carlo approximation. We do so by introducing the Relation-Aware Projecting Direction (RAPD), effectively capturing the pairwise association of each of two pairs of random vectors, each following their ambient law. This enables us to derive the Relation-Aware Slicing Distribution (RASD), a location-scale law corresponding to sampled RAPDs. Finally, we introduce the RASGW distance and its variants, e.g., IWRASGW (Importance Weighted RASGW), which overcome the shortcomings experienced by SGW. We theoretically analyze its properties and substantiate its empirical prowess using extensive experiments on various alignment tasks.

In recent years, contrastive learning has achieved state-of-the-art performance in the territory of self-supervised representation learning. Many previous works have attempted to provide the theoretical understanding underlying the success of contrastive learning. Almost all of them rely on a default assumption, i.e., the label consistency assumption, which may not hold in practice (the probability of failure is called labeling error) due to the strength and randomness of common augmentation strategies, such as random resized crop (RRC). This paper investigates the theoretical impact of labeling error on the downstream classification performance of contrastive learning. We first reveal several significant negative impacts of labeling error on downstream classification risk. To mitigate these impacts, data dimensionality reduction method (e.g., singular value decomposition, SVD) is applied on original data to reduce false positive samples, and establish both theoretical and empirical evaluations. Moreover, it is also found that SVD acts as a double-edged sword, which may lead to the deterioration of downstream classification accuracy due to the reduced connectivity of the augmentation graph. Based on the above observations, we give the augmentation suggestion that we should use some moderate embedding dimension (such as $512, 1024$ in our experiments), data inflation, weak augmentation, and SVD to ensure large graph connectivity and small labeling error to improve model performance.

A new drug usually starts with a tragedy. Born in what is now Zimbabwe, the child of a mechanic and a radiology technician, Ray fled with his family to South Africa during the Zimbabwean War of Liberation. He remembers the journey there in 1980 in a convoy of armored cars. As the sun blazed down, a soldier taught 8-year-old Ray how to fire a machine gun. But his mother kept having to stop.

In the modern era, the complexity and density of data structures have significantly increased, particularly with the advent of technologies such as cloud computing, sensor networks and manifold-based data representations. A notable case within this landscape is the class of measure-valued data, which encompasses data best represented through probability distributions rather than individual observations (Ranjan and Gneiting, 2010; Gneiting and Ranjan, 2013). This framework is prevalent across various fields, including actuarial science, economics and finance, environmental sciences, etc where uncertainty and heterogeneity are inherent and models must reflect the full distributional information. For instance, in economics integrating diverse models allows for the generation of numerous meaningfull probabilistic scenarios that can effectively inform future decision-making (Moral-Benito, 2015; Hong and Martin, 2017; Christensen et al., 2018; Steel, 2020; Koundouri et al., 2024). In environmental sciences, the prediction of future states through stochastic simulation models is crucial for evaluating the consequences of natural hazards (Muis et al., 2015; Hsiang et al., 2017; Fronzek et al., 2022) or improving climatic forecasts (Friederichs and Thorarinsdottir, 2012; Scheuerer and M oller, 2015; Papayiannis et al., 2018).

We explore an ASR-free method for isolated word reading assessment in low-resource settings. Our few-shot approach compares input child speech to a small set of adult-provided reference templates. Inputs and templates are encoded using intermediate layers from large self-supervised learned (SSL) models. Using an Afrikaans child speech benchmark, we investigate design options such as discretising SSL features and barycentre averaging of the templates. Idealised experiments show reasonable performance for adults, but a substantial drop for child speech input, even with child templates. Despite the success of employing SSL representations in low-resource speech tasks, our work highlights the limitations of SSL representations for processing child data when used in a few-shot classification system.