Gaussian Process (GP) models are popular tools in uncertainty quantification (UQ) because they purport to furnish functional uncertainty estimates that can be used to represent model uncertainty . It is often difficult to state with precision what probabilistic interpretation attaches to such an uncertainty, and in what way is it calibrated. Without such a calibration statement, the value of such uncertainty estimates is quite limited and qualitative. We motivate the importance of proper probabilistic calibration of GP predictions by describing how GP predictive calibration failures can cause degraded convergence properties in a target optimization algorithm called T argeted Adaptive Design (T AD). We discuss the interpretation of GP-generated uncertainty intervals in UQ, and how one may learn to trust them, through a formal procedure for covariance kernel validation that exploits the multivariate normal nature of GP predictions. We give simple examples of GP regression misspecified 1-dimensional models, and discuss the situation with respect to higher-dimensional models.

In this work, the Particle Swarm Optimization (PSO) algorithm has been used to train various Variational Quantum Circuits (VQCs). This approach is motivated by the fact that commonly used gradient-based optimization methods can suffer from the barren plateaus problem. PSO is a stochastic optimization technique inspired by the collective behavior of a swarm of birds. The dimension of the swarm, the number of iterations of the algorithm, and the number of trainable parameters can be set. In this study, PSO has been used to train the entire structure of VQCs, allowing it to select which quantum gates to apply, the target qubits, and the rotation angle, in case a rotation is chosen. The algorithm is restricted to choosing from four types of gates: Rx, Ry, Rz, and CNOT. The proposed optimization approach has been tested on various datasets of the MedMNIST, which is a collection of biomedical image datasets designed for image classification tasks. Performance has been compared with the results achieved by classical stochastic gradient descent applied to a predefined VQC. The results show that the PSO can achieve comparable or even better classification accuracy across multiple datasets, despite the PSO using a lower number of quantum gates than the VQC used with gradient descent optimization.

A Large Language Model (LLM) offers versatility across domains and tasks, purportedly benefiting users with a wide variety of behaviors and preferences. We question this perception about an LLM when users have inherently subjective behaviors and preferences, as seen in their ubiquitous and idiosyncratic browsing of websites or apps. The sequential behavior logs of pages, thus generated, form something akin to each user's self-constructed "language", albeit without the structure and grammar imbued in natural languages. We ask: (i) Can a small LM represent the "language of browsing" better than a large LM? (ii) Can an LM with a single set of parameters (or, single LM) adequately capture myriad users' heterogeneous, subjective behaviors and preferences? (iii) Can a single LM with high average performance, yield low variance in performance to make alignment good at user level? We introduce clusterwise LM training, HeTLM (Heterogeneity aware Training of Language Model), appropriate for subjective behaviors. We find that (i) a small LM trained using a page-level tokenizer outperforms large pretrained or finetuned LMs; (ii) HeTLM with heterogeneous cluster specific set of parameters outperforms a single LM of the same family, controlling for the number of parameters; and (iii) a higher mean and a lower variance in generation ensues, implying improved alignment.

The training of high-quality, robust machine learning models for speech-driven 3D facial animation requires a large, diverse dataset of high-quality audio-animation pairs. To overcome the lack of such a dataset, recent work has introduced large pre-trained speech encoders that are robust to variations in the input audio and, therefore, enable the facial animation model to generalize across speakers, audio quality, and languages. However, the resulting facial animation models are prohibitively large and lend themselves only to offline inference on a dedicated machine. In this work, we explore on-device, real-time facial animation models in the context of game development. We overcome the lack of large datasets by using hybrid knowledge distillation with pseudo-labeling. Given a large audio dataset, we employ a high-performing teacher model to train very small student models. In contrast to the pre-trained speech encoders, our student models only consist of convolutional and fully-connected layers, removing the need for attention context or recurrent updates. In our experiments, we demonstrate that we can reduce the memory footprint to up to 3.4 MB and required future audio context to up to 81 ms while maintaining high-quality animations. This paves the way for on-device inference, an important step towards realistic, model-driven digital characters.

Large language models (LLMs) offer an inexpensive yet powerful way to annotate text, but are often inconsistent when compared with experts. These errors can bias downstream estimates of population parameters such as regression coefficients and causal effects. To mitigate this bias, researchers have developed debiasing methods such as Design-based Supervised Learning (DSL) and Prediction-Powered Inference (PPI), which promise valid estimation by combining LLM annotations with a limited number of expensive expert annotations. Although these methods produce consistent estimates under theoretical assumptions, it is unknown how they compare in finite samples of sizes encountered in applied research. We make two contributions. First, we study how each methods performance scales with the number of expert annotations, highlighting regimes where LLM bias or limited expert labels significantly affect results. Second, we compare DSL and PPI across a range of tasks, finding that although both achieve low bias with large datasets, DSL often outperforms PPI on bias reduction and empirical efficiency, but its performance is less consistent across datasets. Our findings indicate that there is a bias-variance tradeoff at the level of debiasing methods, calling for more research on developing metrics for quantifying their efficiency in finite samples.

Recent time-series foundation models exhibit strong abilities to predict physical systems. These abilities include zero-shot forecasting, in which a model forecasts future states of a system given only a short trajectory as context, without knowledge of the underlying physics. Here, we show that foundation models often forecast through a simple parroting strategy, and when they are not parroting they exhibit some shared failure modes such as converging to the mean. As a result, a naive context parroting model that copies directly from the context scores higher than leading time-series foundation models on predicting a diverse range of dynamical systems, including low-dimensional chaos, turbulence, coupled oscillators, and electrocardiograms -- and at a tiny fraction of the computational cost. We draw a parallel between context parroting and induction heads, which explains recent works showing that large language models can often be repurposed for time series forecasting. Our dynamical systems perspective also ties the scaling between forecast accuracy and context length to the fractal dimension of the underlying chaotic attractor, providing insight into previously observed in-context neural scaling laws. By revealing the performance gaps and failure modes of current time-series foundation models, context parroting can guide the design of future foundation models and help identify in-context learning strategies beyond parroting.

--The flipped classroom is a new pedagogical strategy that has been gaining increasing importance recently. Spoken discussion dialog commonly occurs in flipped classroom, which embeds rich information indicating processes and progression of students' learning. This study focuses on learning analytics from spoken discussion dialog in the flipped classroom, which aims to collect and analyze the discussion dialogs in flipped classroom in order to get to know group learning processes and outcomes. We have recently transformed a course using the flipped classroom strategy, where students watched video-recorded lectures at home prior to group-based problem-solving discussions in class. The in-class group discussions were recorded throughout the semester and then transcribed manually. After features are extracted from the dialogs by multiple tools and customized processing techniques, we performed statistical analyses to explore the indicators that are related to the group learning outcomes from face-to-face discussion dialogs in the flipped classroom. Then, machine learning algorithms are applied to the indicators in order to predict the group learning outcome as High, Mid or Low. The best prediction accuracy reaches 78.9%, which demonstrates the feasibility of achieving automatic learning outcome prediction from group discussion dialog in flipped classroom. EARNING analytics is concerned with collection and analyses of data related to learning in order to inform and improve the learning process or their outcomes [1]. Applying properly learning analytics can not only track student progress but also improve student performance [2]. Recent advancements in the development of data science and machine learning techniques has led to a rise in popularity of learning analytics within the educational research field. The flipped classroom is a new pedagogical method, which utilizes asynchronous video lectures and basic practice as homework, and conducts group-based problem solving discussions or activities in the classroom [3]. Since flipped classroom promotes cooperative learning [4, 5] and increases student engagement and motivation [6, 7], it is gaining increasing importance for teaching and learning in recent years. A common in-class activity for the flipped classroom is student group discussions, where participants are involved in solving problems together. Such discussion dialogs embed rich information that cannot be captured objectively by conventional data, such as students' in-class sentiments, degree of concentration, amount of information exchange... etc. Authors are with The Chinese University of Hong Kong, Shatin, N.T., Hong Kong Therefore, spoken discussion dialogs in flipped classroom deserve greater attention for learning analytics, which aims to collect and analyze the discussion dialogs in flipped classroom in order to explore indicators that reflect group learning outcomes.

Atomic data determined by analysis of observed atomic spectra are essential for plasma diagnostics. For each low-ionisation open d- and f-subshell atomic species, around $10^3$ fine structure level energies can be determined through years of analysis of $10^4$ observable spectral lines. We propose the automation of this task by casting the analysis procedure as a Markov decision process and solving it by graph reinforcement learning using reward functions learned on historical human decisions. In our evaluations on existing spectral line lists and theoretical calculations for Co II and Nd II-III, hundreds of level energies were computed within hours, agreeing with published values in 95% of cases for Co II and 54-87% for Nd II-III. As the current efficiency in atomic fine structure determination struggles to meet growing atomic data demands from astronomy and fusion science, our new artificial intelligence approach sets the stage for closing this gap.

The goal of this paper is to accelerate the training of machine learning models, a critical challenge since the training of large-scale deep neural models can be computationally expensive. Stochastic gradient descent (SGD) and its variants are widely used to train deep neural networks. In contrast to traditional approaches that focus on tuning the learning rate, we propose a novel adaptive batch size SGD algorithm, DiveBatch, that dynamically adjusts the batch size. Adapting the batch size is challenging: using large batch sizes is more efficient due to parallel computation, but small-batch training often converges in fewer epochs and generalizes better. To address this challenge, we introduce a data-driven adaptation based on gradient diversity, enabling DiveBatch to maintain the generalization performance of small-batch training while improving convergence speed and computational efficiency. Gradient diversity has a strong theoretical justification: it emerges from the convergence analysis of SGD. Evaluations of DiveBatch on synthetic and CiFar-10, CiFar-100, and Tiny-ImageNet demonstrate that DiveBatch converges significantly faster than standard SGD and AdaBatch (1.06 -- 5.0x), with a slight trade-off in performance.

Vision language models (VLMs) excel in multimodal understanding but are prone to adversarial attacks. Existing defenses often demand costly retraining or significant architecture changes. W e introduce a lightweight defense using tensor decomposition suitable for any pre-trained VLM, requiring no retraining. By decomposing and reconstructing vision encoder representations, it filters adversarial noise while preserving meaning. Experiments with CLIP on COCO and Flickr30K show improved robustness. On Flickr30K, it restores 12.3% performance lost to attacks, raising Recall@1 accuracy from 7.5% to 19.8%. On COCO, it recovers 8.1% performance, improving accuracy from 3.8% to 11.9%. Analysis shows T ensor Train decomposition with low rank (8-32) and low residual strength ( α = 0 . 1 0. 2) is optimal. This method is a practical, plug-and-play solution with minimal overhead for existing VLMs.