搜索结果：handpicked

Financial stress tests based on handpicked scenarios can mislead risk management by overlooking genuinely dangerous configurations or overemphasising shocks that are too implausible to be decision-relevant. We develop a systematic method for generating plausible stress scenarios for financial losses driven by exogenous risk factors. The method exploits a large-deviations principle: conditional on a large loss, the risk factors concentrate near the most likely stress configurations. We use this structure to define representative stress distributions and to extrapolate observed samples into more extreme scenarios while preserving the relative plausibility of stress mechanisms. As a result, the procedure can generate informative stress scenarios even when historical data contain few or no observations in the stressed regime. Numerical experiments on two financial network models show that the method recovers the stressed loss law and key stress diagnostics, including in settings where benchmark generators fail to generate any stressed samples.

Characterisation of structural discontinuity sets in exposed rock faces of underground mine cavities is essential for assessing rock-mass stability, excavation safety, and operational efficiency. UAV and other mobile laser-scanning techniques provide efficient means of collecting point clouds from rock faces. However, the development of a robust and efficient approach for automatic characterisation of discontinuity sets in real-world scenarios, like fully enclosed rock faces in cavities, remains an open research problem. In this study, a new approach is proposed for automatic discontinuity set characterisation that uses a single-shot filtering strategy, an innovative cyclic orientation transformation scheme and a hierarchical clustering technique. The single-shot filtering step isolates planar regions while robustly suppressing noise and high-curvature artefacts in one pass using a signal-processing technique. To address the limitations of Cartesian clustering on polar orientation data, a cyclic orientation transformation scheme is developed, enabling accurate representation of dip angle and dip direction in Cartesian space. The transformed orientations are then characterised into

Predicting a mobile app's popularity before its first release can provide developers with a strategic advantage in a competitive marketplace, yet it remains a challenging problem. This study explores the extent to which internal software metrics, measurable from source code before deployment, can predict an app's popularity (i.e., ratings and downloads per year) at inception. For our analysis, we constructed a rigorously filtered dataset of 446 open-source Java-based Android apps that are available on both F-Droid and Google Play Store. Using app source code from F-Droid, we extracted a wide array of internal metrics, including system-, class-, and method-level code metrics, code smells, and app metadata. Popularity-related information, including reviews and download counts, was collected from the Play Store. We evaluate regression and classification models across three feature sets: a minimal Size-only baseline, a domain-informed Handpicked set, and a Voting set derived via feature selection algorithms. Our results show that, for both app ratings and number of downloads, regression models perform poorly due to skewed rating distributions and a highly scattered range of download co

When selecting data for training large-scale models, standard practice is to filter for examples that match human notions of data quality. Such filtering yields qualitatively clean datapoints that intuitively should improve model behavior. However, in practice the opposite can often happen: we find that selecting according to similarity with "high quality" data sources may not increase (and can even hurt) performance compared to randomly selecting data. To develop better methods for selecting data, we start by framing dataset selection as an optimization problem that we can directly solve for: given target tasks, a learning algorithm, and candidate data, select the subset that maximizes model performance. This framework thus avoids handpicked notions of data quality, and instead models explicitly how the learning process uses train datapoints to predict on the target tasks. Our resulting method greatly improves language model (LM) performance on both pre-specified tasks and previously unseen tasks. Specifically, choosing target tasks representative of standard LM problems and evaluating on diverse held-out benchmarks, our selected datasets provide a 2x compute multiplier over basel

Building on previous research on frequency allocation optimization for superconducting circuit quantum processors, this work incorporates several new techniques to improve overall solution quality. New features include tightening constraints, imposing edgewise differences, including edge orientation in the optimization, and integrating multimodule designs with various boundary conditions. These enhancements allow for greater flexibility in processor design by eliminating the need for handpicked orientations. We support the efficient assembly of large processors with dense connectivity by choosing the best boundary conditions. Examples demonstrate that, at low computational cost, the new optimization approach finds a frequency configuration for a square chip with over 1,000 qubits and over 10% yield at much larger dispersion levels than required by previous approaches.

Clifford Group Equivariant Neural Networks (CGENNs) leverage Clifford algebras and multivectors as an alternative approach to incorporating group equivariance to ensure symmetry constraints in neural representations. In principle, this formulation generalizes to orthogonal groups and preserves equivariance regardless of the metric signature. However, previous works have restricted internal network representations to Euclidean or Minkowski (pseudo-)metrics, handpicked depending on the problem at hand. In this work, we propose an alternative method that enables the metric to be learned in a data-driven fashion, allowing the CGENN network to learn more flexible representations. Specifically, we populate metric matrices fully, ensuring they are symmetric by construction, and leverage eigenvalue decomposition to integrate this additional learnable component into the original CGENN formulation in a principled manner. Additionally, we motivate our method using insights from category theory, which enables us to explain Clifford algebras as a categorical construction and guarantee the mathematical soundness of our approach. We validate our method in various tasks and showcase the advantages

This paper investigates cash productivity as a signal for future stock performance, building on the cash-return framework of Faulkender and Wang (2006). Using financial and market data from WRDS, we calculate cash returns as a proxy for operational efficiency and evaluate a long-only strategy applied to Nasdaq-listed non-financial firms. Results show limited predictive power across the broader Nasdaq universe but strong performance in a handpicked portfolio, which achieves significant positive alpha after controlling for the Fama-French three factors. These findings underscore the importance of refined universe selection. While promising, the strategy requires further validation, including the incorporation of transaction costs and performance testing across economic cycles. Our results suggest that cash productivity, when combined with other complementary signals and careful universe selection, can be a valuable tool for generating excess returns.

Chart understanding plays a pivotal role when applying Multimodal Large Language Models (MLLMs) to real-world tasks such as analyzing scientific papers or financial reports. However, existing datasets often focus on oversimplified and homogeneous charts with template-based questions, leading to an over-optimistic measure of progress. We demonstrate that although open-source models can appear to outperform strong proprietary models on these benchmarks, a simple stress test with slightly different charts or questions can deteriorate performance by up to 34.5%. In this work, we propose CharXiv, a comprehensive evaluation suite involving 2,323 natural, challenging, and diverse charts from arXiv papers. CharXiv includes two types of questions: 1) descriptive questions about examining basic chart elements and 2) reasoning questions that require synthesizing information across complex visual elements in the chart. To ensure quality, all charts and questions are handpicked, curated, and verified by human experts. Our results reveal a substantial, previously underestimated gap between the reasoning skills of the strongest proprietary model (i.e., GPT-4o), which achieves 47.1% accuracy, and

Bioacoustic data from Passive Acoustic Monitoring (PAM) generates large datasets where obtaining detailed auditing and labelling is often impractical, resulting in weak annotations (e.g., presence/absence of species over several minutes of recording). In order to effectively capture the complex temporal patterns and key features of long audio segments, we propose a framework comprising dataset standardisation, feature extraction, and classification via Temporal Convolutional Networks (TCN). This approach eliminates the necessity for setting heuristic decision rules or creating time-consuming strong labels. To demonstrate the effectiveness of our approach, we use sperm whale (\textit{Physeter macrocephalus}) click trains in 4-minute recordings as a case study, from a dataset comprising diverse sources and deployment conditions to maximise generalisability. Our TCN classifiers achieve recall rates exceeding 0.83 at a 0.13 false positive rate, comparable to agreement rates between expert annotators. We compare two methods of feature extraction, Variational AutoEncoders (VAEs) and traditional handpicking of features, and found them to yield similar performance results, with the VAE-bas

We present unexpected findings from a large-scale benchmark study evaluating Conditional Average Treatment Effect (CATE) estimation algorithms, i.e., CATE models. By running 16 modern CATE models on 12 datasets and 43,200 sampled variants generated through diverse observational sampling strategies, we find that: (a) 62\% of CATE estimates have a higher Mean Squared Error (MSE) than a trivial zero-effect predictor, rendering them ineffective; (b) in datasets with at least one useful CATE estimate, 80\% still have higher MSE than a constant-effect model; and (c) Orthogonality-based models outperform other models only 30\% of the time, despite widespread optimism about their performance. These findings highlight significant challenges in current CATE models and underscore the need for broader evaluation and methodological improvements. Our findings stem from a novel application of \textit{observational sampling}, originally developed to evaluate Average Treatment Effect (ATE) estimates from observational methods with experiment data. To adapt observational sampling for CATE evaluation, we introduce a statistical parameter, $Q$, equal to MSE minus a constant and preserves the ranking o

Several enhanced sampling techniques rely on the definition of collective variables to effectively explore free energy landscapes. Existing variables that describe the progression along a reactive pathway offer an elegant solution but face a number of limitations. In this paper, we address these challenges by introducing a new path-like collective variable called the `Deep-locally-non-linear-embedding', which is inspired by principles of the locally linear embedding technique and is trained on a reactive trajectory. The variable mimics the ideal reaction coordinate by automatically generating a non-linear combination of features through a differentiable generalized autoencoder that combines a neural network with a continuous k-nearest-neighbor selection. Among the key advantages of this method is its capability to automatically choose the metric for searching neighbors and to learn the path from state A to state B without the need to handpick landmarks a priori. We demonstrate the effectiveness of DeepLNE by showing that the progression along the path variable closely approximates the ideal reaction coordinate in toy models such as the Müller-Brown-potential and alanine dipeptide.

The observation of very-high-energy (VHE, E>100 GeV) gamma rays is mediated by the imaging atmospheric Cherenkov technique (IACTs). At these energies, gamma rays interact with the atmosphere to create a cascade of electromagnetic air showers that are visible to the IACT cameras on the ground with distinct morphological and temporal features. However, hadrons with significantly higher incidence rates are also imaged with similar features, and must be distinguished with handpicked parameters extracted from the images. The advent of sophisticated deep learning models has enabled an alternative image analysis technique that has been shown to improve the detection of gamma rays, by improving background rejection. In this study, we propose an unsupervised Wasserstein Generative Adversarial Network (WGAN) framework trained on normalized, uncleaned stereoscopic shower images of real events from the VERITAS observatory to extract the landscape of their latent space and optimize against the corresponding inferred latent space of simulated gamma-ray events. We aim to develop a data driven approach to guide the understanding of the extracted features of real gamma-ray images, and will optim

With the rise of digital media content production, the need for analyzing movies and TV series episodes to locate the main cast of characters precisely is gaining importance.Specifically, Video Face Clustering aims to group together detected video face tracks with common facial identities. This problem is very challenging due to the large range of pose, expression, appearance, and lighting variations of a given face across video frames. Generic pre-trained Face Identification (ID) models fail to adapt well to the video production domain, given its high dynamic range content and also unique cinematic style. Furthermore, traditional clustering algorithms depend on hyperparameters requiring individual tuning across datasets. In this paper, we present a novel video face clustering approach that learns to adapt a generic face ID model to new video face tracks in a fully self-supervised fashion. We also propose a parameter-free clustering algorithm that is capable of automatically adapting to the finetuned model's embedding space for any input video. Due to the lack of comprehensive movie face clustering benchmarks, we also present a first-of-kind movie dataset: MovieFaceCluster. Our dat

Edit fidelity is a significant issue in open-world controllable generative image editing. Recently, CLIP-based approaches have traded off simplicity to alleviate these problems by introducing spatial attention in a handpicked layer of a StyleGAN. In this paper, we propose CoralStyleCLIP, which incorporates a multi-layer attention-guided blending strategy in the feature space of StyleGAN2 for obtaining high-fidelity edits. We propose multiple forms of our co-optimized region and layer selection strategy to demonstrate the variation of time complexity with the quality of edits over different architectural intricacies while preserving simplicity. We conduct extensive experimental analysis and benchmark our method against state-of-the-art CLIP-based methods. Our findings suggest that CoralStyleCLIP results in high-quality edits while preserving the ease of use.

Highly specific datasets of scientific literature are important for both research and education. However, it is difficult to build such datasets at scale. A common approach is to build these datasets reductively by applying topic modeling on an established corpus and selecting specific topics. A more robust but time-consuming approach is to build the dataset constructively in which a subject matter expert (SME) handpicks documents. This method does not scale and is prone to error as the dataset grows. Here we showcase a new tool, based on machine learning, for constructively generating targeted datasets of scientific literature. Given a small initial "core" corpus of papers, we build a citation network of documents. At each step of the citation network, we generate text embeddings and visualize the embeddings through dimensionality reduction. Papers are kept in the dataset if they are "similar" to the core or are otherwise pruned through human-in-the-loop selection. Additional insight into the papers is gained through sub-topic modeling using SeNMFk. We demonstrate our new tool for literature review by applying it to two different fields in machine learning.

Entity Matching is the task of deciding if two entity descriptions refer to the same real-world entity. State-of-the-art entity matching methods often rely on fine-tuning Transformer models such as BERT or RoBERTa. Two major drawbacks of using these models for entity matching are that (i) the models require significant amounts of fine-tuning data for reaching a good performance and (ii) the fine-tuned models are not robust concerning out-of-distribution entities. In this paper, we investigate using ChatGPT for entity matching as a more robust, training data-efficient alternative to traditional Transformer models. We perform experiments along three dimensions: (i) general prompt design, (ii) in-context learning, and (iii) provision of higher-level matching knowledge. We show that ChatGPT is competitive with a fine-tuned RoBERTa model, reaching a zero-shot performance of 82.35% F1 on a challenging matching task on which RoBERTa requires 2000 training examples for reaching a similar performance. Adding in-context demonstrations to the prompts further improves the F1 by up to 7.85% when using similarity-based example selection. Always using the same set of 10 handpicked demonstrations

Hypothetical induction is recognized as the main reasoning type when scientists make observations about the world and try to propose hypotheses to explain those observations. Past research on hypothetical induction is under a constrained setting: (1) the observation annotations in the dataset are carefully manually handpicked sentences (resulting in a close-domain setting); and (2) the ground truth hypotheses are mostly commonsense knowledge, making the task less challenging. In this work, we tackle these problems by proposing the first dataset for social science academic hypotheses discovery, with the final goal to create systems that automatically generate valid, novel, and helpful scientific hypotheses, given only a pile of raw web corpus. Unlike previous settings, the new dataset requires (1) using open-domain data (raw web corpus) as observations; and (2) proposing hypotheses even new to humanity. A multi-module framework is developed for the task, including three different feedback mechanisms to boost performance, which exhibits superior performance in terms of both GPT-4 based and expert-based evaluation. To the best of our knowledge, this is the first work showing that LLMs

The performance of convolutional neural networks (CNN) depends heavily on their architectures. Transfer learning performance of a CNN relies quite strongly on selection of its trainable layers. Selecting the most effective update layers for a certain target dataset often requires expert knowledge on CNN architecture which many practitioners do not posses. General users prefer to use an available architecture (e.g. GoogleNet, ResNet, EfficientNet etc.) that is developed by domain experts. With the ever-growing number of layers, it is increasingly becoming quite difficult and cumbersome to handpick the update layers. Therefore, in this paper we explore the application of genetic algorithm to mitigate this problem. The convolutional layers of popular pretrained networks are often grouped into modules that constitute their building blocks. We devise a genetic algorithm to select blocks of layers for updating the parameters. By experimenting with EfficientNetB0 pre-trained on ImageNet and using Food-101, CIFAR-100 and MangoLeafBD as target datasets, we show that our algorithm yields similar or better results than the baseline in terms of accuracy, and requires lower training and evaluat

This paper develops an optimal data aggregation policy for learning-based traffic control systems based on imagery collected from Road Side Units (RSUs) under imperfect communications. Our focus is optimizing semantic information flow from RSUs to a nearby edge server or cloud-based processing units by maximizing data diversity based on the target machine learning application while taking into account heterogeneous channel conditions (e.g., delay, error rate) and constrained total transmission rate. As a proof-of-concept, we enforce fairness among class labels to increase data diversity for classification problems. The developed constrained optimization problem is non-convex. Hence it does not admit a closed-form solution, and the exhaustive search is NP-hard in the number of RSUs. To this end, we propose an approximate algorithm that applies a greedy interval-by-interval scheduling policy by selecting RSUs to transmit. We use coalition game formulation to maximize the overall added fairness by the selected RSUs in each transmission interval. Once, RSUs are selected, we employ a maximum uncertainty method to handpick data samples that contribute the most to the learning performance