In this paper, we experimented with the SpeechT5 model pre-trained on large-scale datasets. We pre-trained the foundation model from scratch and fine-tuned it on a large-scale robust multi-speaker text-to-speech (TTS) task. We tested the model capabilities in a zero- and few-shot scenario. Based on two listening tests, we evaluated the synthetic audio quality and the similarity of how synthetic voices resemble real voices. Our results showed that the SpeechT5 model can generate a synthetic voice for any speaker using only one minute of the target speaker's data. We successfully demonstrated the high quality and similarity of our synthetic voices on publicly known Czech politicians and celebrities.

Extracting social determinants of health (SDoH) from unstructured medical notes depends heavily on labor-intensive annotations, which are typically task-specific, hampering reusability and limiting sharing. In this study we introduced SDoH-GPT, a simple and effective few-shot Large Language Model (LLM) method leveraging contrastive examples and concise instructions to extract SDoH without relying on extensive medical annotations or costly human intervention. It achieved tenfold and twentyfold reductions in time and cost respectively, and superior consistency with human annotators measured by Cohen's kappa of up to 0.92. The innovative combination of SDoH-GPT and XGBoost leverages the strengths of both, ensuring high accuracy and computational efficiency while consistently maintaining 0.90+ AUROC scores. Testing across three distinct datasets has confirmed its robustness and accuracy. This study highlights the potential of leveraging LLMs to revolutionize medical note classification, demonstrating their capability to achieve highly accurate classifications with significantly reduced time and cost.

Phase recognition in surgical videos is crucial for enhancing computer-aided surgical systems as it enables automated understanding of sequential procedural stages. Existing methods often rely on fixed temporal windows for video analysis to identify dynamic surgical phases. Thus, they struggle to simultaneously capture short-, mid-, and long-term information necessary to fully understand complex surgical procedures. To address these issues, we propose Multi-Scale Transformers for Surgical Phase Recognition (MuST), a novel Transformer-based approach that combines a Multi-Term Frame encoder with a Temporal Consistency Module to capture information across multiple temporal scales of a surgical video. Our Multi-Term Frame Encoder computes interdependencies across a hierarchy of temporal scales by sampling sequences at increasing strides around the frame of interest. Furthermore, we employ a long-term Transformer encoder over the frame embeddings to further enhance long-term reasoning. MuST achieves higher performance than previous state-of-the-art methods on three different public benchmarks.

The signature of a path is a sequence of integrals, applied iteratively to the components of the path, which allows it to describe the path precisely and to summarise its characteristics, especially the geometrical ones. This concept emerged in the 1950s and was originally studied by K. T. Chen, who developed the theory and gave the first significant results that justified the interest in the signature. After that, the signature formed part of the Terry Lyons' theory of rough paths, which is key in the field of stochastic calculus and in the study of differential equations controlled by rough paths. Thanks to the development of this theory, the signature was generalised to apply to certain paths of finite variation and regained some relevance. More recently, applications of the signature have also been found in the field of machine learning, where its properties for describing paths are useful for summarising the data sequences used in this discipline and for revealing their properties, which facilitates the training of models that have to make decisions based on the data. Before we begin, we start by defining some concepts.

Humans and many animals show remarkably adaptive behavior and can respond differently to the same input depending on their internal goals. The brain not only represents the intermediate abstractions needed to perform a computation but also actively maintains a representation of the computation itself (task abstraction). Such separation of the computation and its abstraction is associated with faster learning, flexible decision-making, and broad generalization capacity. We investigate if such benefits might extend to neural networks trained with task abstractions. For such benefits to emerge, one needs a task inference mechanism that possesses two crucial abilities: First, the ability to infer abstract task representations when no longer explicitly provided (task inference), and second, manipulate task representations to adapt to novel problems (task recomposition). To tackle this, we cast task inference as an optimization problem from a variational inference perspective and ground our approach in an expectation-maximization framework. We show that gradients backpropagated through a neural network to a task representation layer are an efficient heuristic to infer current task demands, a process we refer to as gradient-based inference (GBI). Further iterative optimization of the task representation layer allows for recomposing abstractions to adapt to novel situations. Using a toy example, a novel image classifier, and a language model, we demonstrate that GBI provides higher learning efficiency and generalization to novel tasks and limits forgetting. Moreover, we show that GBI has unique advantages such as preserving information for uncertainty estimation and detecting out-of-distribution samples.

While large language models are rapidly moving towards consumer-facing applications, they are often still prone to factual errors and hallucinations. In order to reduce the potential harms that may come from these errors, it is important for users to know to what extent they can trust an LLM when it makes a factual claim. To this end, we study the problem of uncertainty quantification of factual correctness in long-form natural language generation. Given some output from a large language model, we study both uncertainty at the level of individual claims contained within the output (via calibration) and uncertainty across the entire output itself (via conformal prediction). Moreover, we invoke multicalibration and multivalid conformal prediction to ensure that such uncertainty guarantees are valid both marginally and across distinct groups of prompts. Using the task of biography generation, we demonstrate empirically that having access to and making use of additional group attributes for each prompt improves both overall and group-wise performance. As the problems of calibration, conformal prediction, and their multi-group counterparts have not been extensively explored previously in the context of long-form text generation, we consider these empirical results to form a benchmark for this setting.

In this work, we consider the non-invasive medical imaging modality of Electrical Impedance Tomography, where the problem is to recover the conductivity in a medium from a set of data that arises out of a current-to-voltage map (Neumann-to-Dirichlet operator) defined on the boundary of the medium. We formulate this inverse problem as an operator-learning problem where the goal is to learn the implicitly defined operator-to-function map between the space of Neumann-to-Dirichlet operators to the space of admissible conductivities. Subsequently, we use an operator-learning architecture, popularly called DeepONets, to learn this operator-to-function map. Thus far, most of the operator learning architectures have been implemented to learn operators between function spaces. In this work, we generalize the earlier works and use a DeepONet to actually learn an operator-to-function map. We provide a Universal Approximation Theorem type result which guarantees that this implicitly defined operator-to-function map between the the space of Neumann-to-Dirichlet operator to the space of conductivity function can be approximated to an arbitrary degree using such a DeepONet. Furthermore, we provide a computational implementation of our proposed approach and compare it against a standard baseline. We show that the proposed approach achieves good reconstructions and outperforms the baseline method in our experiments.

To reduce the experimental validation workload for chemical researchers and accelerate the design and optimization of high-energy-density lithium metal batteries, we aim to leverage models to automatically predict Coulombic Efficiency (CE) based on the composition of liquid electrolytes. There are mainly two representative paradigms in existing methods: machine learning and deep learning. However, the former requires intelligent input feature selection and reliable computational methods, leading to error propagation from feature estimation to model prediction, while the latter (e.g. MultiModal-MoLFormer) faces challenges of poor predictive performance and overfitting due to limited diversity in augmented data. To tackle these issues, we propose a novel method COEFF (COlumbic EFficiency prediction via Fine-tuned models), which consists of two stages: pre-training a chemical general model and fine-tuning on downstream domain data. Firstly, we adopt the publicly available MoLFormer model to obtain feature vectors for each solvent and salt in the electrolyte. Then, we perform a weighted average of embeddings for each token across all molecules, with weights determined by the respective electrolyte component ratios. Finally, we input the obtained electrolyte features into a Multi-layer Perceptron or Kolmogorov-Arnold Network to predict CE. Experimental results on a real-world dataset demonstrate that our method achieves SOTA for predicting CE compared to all baselines. Data and code used in this work will be made publicly available after the paper is published.

In semiconductor manufacturing, the wafer dicing process is central yet vulnerable to defects that significantly impair yield - the proportion of defect-free chips. Deep neural networks are the current state of the art in (semi-)automated visual inspection. However, they are notoriously known to require a particularly large amount of data for model training. To address these challenges, we explore the application of generative adversarial networks (GAN) for image data augmentation and classification of semiconductor wafer dicing induced defects to enhance the variety and balance of training data for visual inspection systems. With this approach, synthetic yet realistic images are generated that mimic real-world dicing defects. We employ three different GAN variants for high-resolution image synthesis: Deep Convolutional GAN (DCGAN), CycleGAN, and StyleGAN3. Our work-in-progress results demonstrate that improved classification accuracies can be obtained, showing an average improvement of up to 23.1 % from 65.1 % (baseline experiment) to 88.2 % (DCGAN experiment) in balanced accuracy, which may enable yield optimization in production.

We propose a deep learning framework, DL-opt, designed to efficiently solve for optimal policies in quantifiable general equilibrium trade models. DL-opt integrates (i) a nested fixed point (NFXP) formulation of the optimization problem, (ii) automatic implicit differentiation to enhance gradient descent for solving unilateral optimal policies, and (iii) a best-response dynamics approach for finding Nash equilibria. Utilizing DL-opt, we solve for non-cooperative tariffs and industrial subsidies across 7 economies and 44 sectors, incorporating sectoral external economies of scale. Our quantitative analysis reveals significant sectoral heterogeneity in Nash policies: Nash industrial subsidies increase with scale elasticities, whereas Nash tariffs decrease with trade elasticities. Moreover, we show that global dual competition, involving both tariffs and industrial subsidies, results in lower tariffs and higher welfare outcomes compared to a global tariff war.