共找到 20 条结果
We introduce Padyam2Gadyam a dataset for the task of poem-to-prose translation from 13th-17th Century Telugu Classical Poetry to contemporary Telugu and English prose. The dataset consists of 600 poems and their human-verified Telugu and English prose translations. We evaluated 2 machine translation systems and 5 contemporary Large Language Models (LLMs) on their ability to do poem-to-prose translation into Telugu and English using this dataset. Our results indicate that while the general purpose LLMs are better than the machine translation systems for this task, there are systematic issues with the generation and evaluation of prose translation in both languages across systems.
This article explores the potential of generative AI (GenAI) to support actuarial practice through four implemented case studies. It situates these case studies within the broader evolution of artificial intelligence in actuarial science, from early neural networks and machine learning to modern transformer-based GenAI systems. The first case study illustrates how large language models (LLMs) can improve claim cost prediction by extracting informative features from unstructured text for use in the underlying supervised learning task. The second case study demonstrates the automation of market comparisons using Retrieval-Augmented Generation to identify, extract, and structure relevant information from insurers' annual reports. The third case study highlights the capabilities of fine-tuned vision-enabled LLMs in classifying car damage types and extracting contextual information from images. The fourth case study presents a multi-agent system that autonomously migrates actuarial legacy code from R to Python and validates the translation against the original code's outputs. In addition to these case studies, we outline further GenAI applications in the insurance industry. Finally, we
In the short period since the release of ChatGPT, large language models (LLMs) have changed the software engineering research landscape. While there are numerous opportunities to use LLMs for supporting research or software engineering tasks, solid science needs rigorous empirical evaluations. However, so far, there are no specific guidelines for conducting and assessing studies involving LLMs in software engineering research. Our focus is on empirical studies that either use LLMs as part of the research process or studies that evaluate existing or new tools that are based on LLMs. This paper contributes the first set of holistic guidelines for such studies. Our goal is to start a discussion in the software engineering research community to reach a common understanding of our standards for high-quality empirical studies involving LLMs.
We present a mechanistic analysis of literary style in GPT-2, identifying individual neurons that discriminate between exemplary prose and rigid AI-generated text. Using Herman Melville's Bartleby, the Scrivener as a corpus, we extract activation patterns from 355 million parameters across 32,768 neurons in late layers. We find 27,122 statistically significant discriminative neurons ($p < 0.05$), with effect sizes up to $|d| = 1.4$. Through systematic ablation studies, we discover a paradoxical result: while these neurons correlate with literary text during analysis, removing them often improves rather than degrades generated prose quality. Specifically, ablating 50 high-discriminating neurons yields a 25.7% improvement in literary style metrics. This demonstrates a critical gap between observational correlation and causal necessity in neural networks. Our findings challenge the assumption that neurons which activate on desirable inputs will produce those outputs during generation, with implications for mechanistic interpretability research and AI alignment.
Large language models produce em dashes at varying rates, and the observation that some models "overuse" them has become one of the most widely discussed markers of AI-generated text. Yet no mechanistic account of this pattern exists, and the parallel observation that LLMs default to markdown-formatted output has never been connected to it. We propose that the em dash is markdown leaking into prose -- the smallest surviving unit of the structural orientation that LLMs acquire from markdown-saturated training corpora. We present a five-step genealogy connecting training data composition, structural internalization, the dual-register status of the em dash, and post-training amplification. We test this with a two-condition suppression experiment across twelve models from five providers (Anthropic, OpenAI, Meta, Google, DeepSeek): when models are instructed to avoid markdown formatting, overt features (headers, bullets, bold) are eliminated or nearly eliminated, but em dashes persist -- except in Meta's Llama models, which produce none at all. Em dash frequency and suppression resistance vary from 0.0 per 1,000 words (Llama) to 9.1 (GPT-4.1 under suppression), functioning as a signatur
Registering two captures of the same indoor space taken at different times underpins persistent spatial memory for robots and AR systems, yet the realistic version of this task is egocentric and its most scalable form is RGB-only. Head-mounted cameras yield blurry, fast-moving, partially overlapping views from which dense geometry is hard to recover. Classical registration leans on exactly the clean point clouds this setting lacks, while learned scene-graph methods require a pre-built or annotated graph and a trained matcher that we find brittle under egocentric data. We take a different route, using a pretrained vision-language model as the source of both scene understanding and cross-scan matching. Our method, PROSE (Prompted Scene rEgistration), lifts each RGB sequence into an object-level 3D scene graph using off-the-shelf foundation models for geometry, segmentation, and language, then prompts the same VLM to match object instances across the two RGB sequences. To make this matching tractable and reliable, we leverage object heights as a prior and verify each proposed match with a paired same/different query, then solve for the rigid transform by hypothesizing a candidate per
Large Language Models (LLMs) are increasingly treated as universal, general-purpose solutions across NLP tasks, particularly in English. But does this assumption hold for low-resource, morphologically rich languages such as Sanskrit? We address this question by comparing instruction-tuned and in-context-prompted LLMs with smaller task-specific encoder-decoder models on the Sanskrit poetry-to-prose conversion task. This task is intrinsically challenging: Sanskrit verse exhibits free word order combined with rigid metrical constraints, and its conversion to canonical prose (anvaya) requires multi-step reasoning involving compound segmentation, dependency resolution, and syntactic linearisation. This makes it an ideal testbed to evaluate whether LLMs can surpass specialised models. For LLMs, we apply instruction fine-tuning on general-purpose models and design in-context learning templates grounded in Paninian grammar and classical commentary heuristics. For task-specific modelling, we fully fine-tune a ByT5-Sanskrit Seq2Seq model. Our experiments show that domain-specific fine-tuning of ByT5-Sanskrit significantly outperforms all instruction-driven LLM approaches. Human evaluation st
[Background] Systematic literature reviews (SLRs) are essential for synthesizing evidence in Software Engineering (SE), but keeping them up-to-date requires substantial effort. Study selection, one of the most labor-intensive steps, involves reviewing numerous studies and requires multiple reviewers to minimize bias and avoid loss of evidence. [Objective] This study aims to evaluate if Machine Learning (ML) text classification models can support reviewers in the study selection for SLR updates. [Method] We reproduce the study selection of an SLR update performed by three SE researchers. We trained two supervised ML models (Random Forest and Support Vector Machines) with different configurations using data from the original SLR. We calculated the study selection effectiveness of the ML models for the SLR update in terms of precision, recall, and F-measure. We also compared the performance of human-ML pairs with human-only pairs when selecting studies. [Results] The ML models achieved a modest F-score of 0.33, which is insufficient for reliable automation. However, we found that such models can reduce the study selection effort by 33.9% without loss of evidence (keeping a 100% recall
Accurate and contextually faithful responses are critical when applying large language models (LLMs) to sensitive and domain-specific tasks, such as answering queries related to quranic studies. General-purpose LLMs often struggle with hallucinations, where generated responses deviate from authoritative sources, raising concerns about their reliability in religious contexts. This challenge highlights the need for systems that can integrate domain-specific knowledge while maintaining response accuracy, relevance, and faithfulness. In this study, we investigate 13 open-source LLMs categorized into large (e.g., Llama3:70b, Gemma2:27b, QwQ:32b), medium (e.g., Gemma2:9b, Llama3:8b), and small (e.g., Llama3.2:3b, Phi3:3.8b). A Retrieval-Augmented Generation (RAG) is used to make up for the problems that come with using separate models. This research utilizes a descriptive dataset of Quranic surahs including the meanings, historical context, and qualities of the 114 surahs, allowing the model to gather relevant knowledge before responding. The models are evaluated using three key metrics set by human evaluators: context relevance, answer faithfulness, and answer relevance. The findings re
Speech enhancement (SE) is the foundational task of enhancing the clarity and quality of speech in the presence of non-stationary additive noise. While deterministic deep learning models have been commonly employed for SE, recent research indicates that generative models, such as denoising diffusion probabilistic models (DDPMs), have shown promise. However, unlike speech generation, SE has a strong constraint in generating results in accordance with the underlying ground-truth signal. Additionally, for a wide variety of applications, SE systems need to be employed in real-time, and traditional diffusion models (DMs) requiring many iterations of a large model during inference are inefficient. To address these issues, we propose ProSE (diffusion-based Priors for SE), a novel methodology based on an alternative framework for applying diffusion models to SE. Specifically, we first apply DDPMs to generate priors in a latent space due to their powerful distribution mapping capabilities. The priors are then integrated into a transformer-based regression model for SE. The priors guide the regression model in the enhancement process. Since the diffusion process is applied to a compact laten
Comics have long been a popular form of storytelling, offering visually engaging narratives that captivate audiences worldwide. However, the visual nature of comics presents a significant barrier for visually impaired readers, limiting their access to these engaging stories. In this work, we provide a pragmatic solution to this accessibility challenge by developing an automated system that generates text-based literary narratives from manga comics. Our approach aims to create an evocative and immersive prose that not only conveys the original narrative but also captures the depth and complexity of characters, their interactions, and the vivid settings in which they reside. To this end we make the following contributions: (1) We present a unified model, Magiv3, that excels at various functional tasks pertaining to comic understanding, such as localising panels, characters, texts, and speech-bubble tails, performing OCR, grounding characters etc. (2) We release human-annotated captions for over 3300 Japanese comic panels, along with character grounding annotations, and benchmark large vision-language models in their ability to understand comic images. (3) Finally, we demonstrate how
We propose PROSE-FD, a zero-shot multimodal PDE foundational model for simultaneous prediction of heterogeneous two-dimensional physical systems related to distinct fluid dynamics settings. These systems include shallow water equations and the Navier-Stokes equations with incompressible and compressible flow, regular and complex geometries, and different buoyancy settings. This work presents a new transformer-based multi-operator learning approach that fuses symbolic information to perform operator-based data prediction, i.e. non-autoregressive. By incorporating multiple modalities in the inputs, the PDE foundation model builds in a pathway for including mathematical descriptions of the physical behavior. We pre-train our foundation model on 6 parametric families of equations collected from 13 datasets, including over 60K trajectories. Our model outperforms popular operator learning, computer vision, and multi-physics models, in benchmark forward prediction tasks. We test our architecture choices with ablation studies.
This thesis presents Abstractive Text Summarization models for contemporary Sanskrit prose. The first chapter, titled Introduction, presents the motivation behind this work, the research questions, and the conceptual framework. Sanskrit is a low-resource inflectional language. The key research question that this thesis investigates is what the challenges in developing an abstractive TS for Sanskrit. To answer the key research questions, sub-questions based on four different themes have been posed in this work. The second chapter, Literature Review, surveys the previous works done. The third chapter, data preparation, answers the remaining three questions from the third theme. It reports the data collection and preprocessing challenges for both language model and summarization model trainings. The fourth chapter reports the training and inference of models and the results obtained therein. This research has initiated a pipeline for Sanskrit abstractive text summarization and has reported the challenges faced at every stage of the development. The research questions based on every theme have been answered to answer the key research question.
Die studies are fundamental to quantifying ancient monetary production, providing insights into the relationship between coinage, politics, and history. The process requires tedious manual work, which limits the size of the corpora that can be studied. Few works have attempted to automate this task, and none have been properly released and evaluated from a computer vision perspective. We propose a fully automatic approach that introduces several innovations compared to previous methods. We rely on fast and robust local descriptors matching that is set automatically. Second, the core of our proposal is a clustering-based approach that uses an intrinsic metric (that does not need the ground truth labels) to determine its critical hyper-parameters. We validate the approach on two corpora of Greek coins, propose an automatic implementation and evaluation of previous baselines, and show that our approach significantly outperforms them.
Languages such as P4 and NPL have enabled a wide and diverse range of networking applications that take advantage of programmable dataplanes. However, software development in these languages is difficult. To address this issue, high-level languages have been designed to offer programmers powerful abstractions that reduce the time, effort and domain-knowledge required for developing networking applications. These languages are then translated by a compiler into P4/NPL code. Inspired by the recent success of Large Language Models (LLMs) in the task of code generation, we propose to raise the level of abstraction even higher, employing LLMs to translate prose into high-level networking code. We analyze the problem, focusing on the motivation and opportunities, as well as the challenges involved and sketch out a roadmap for the development of a system that can generate high-level dataplane code from natural language instructions. We present some promising preliminary results on generating Lucid code from natural language.
Software engineering (SE) is full of abstract concepts that are crucial for both researchers and practitioners, such as programming experience, team productivity, code comprehension, and system security. Secondary studies aimed at summarizing research on the influences and consequences of such concepts would therefore be of great value. However, the inability to measure abstract concepts directly poses a challenge for secondary studies: primary studies in SE can operationalize such concepts in many ways. Standardized measurement instruments are rarely available, and even if they are, many researchers do not use them or do not even provide a definition for the studied concept. SE researchers conducting secondary studies therefore have to decide a) which primary studies intended to measure the same construct, and b) how to compare and aggregate vastly different measurements for the same construct. In this experience report, we discuss the challenge of study selection in SE secondary research on latent variables. We report on two instances where we found it particularly challenging to decide which primary studies should be included for comparison and synthesis, so as not to end up com
Empirical studies form an integral part of visualization research. Not only can they facilitate the evaluation of various designs, techniques, systems, and practices in visualization, but they can also enable the discovery of the causalities explaining why and how visualization works. This state-of-the-art report focuses on controlled and semi-controlled empirical studies conducted in laboratories and crowd-sourcing environments. In particular, the survey provides a taxonomic analysis of over 129 empirical studies in the visualization literature. It juxtaposes these studies with topic developments between 1978 and 2017 in psychology, where controlled empirical studies have played a predominant role in research. To help appreciate this broad context, the paper provides two case studies in detail, where specific visualization-related topics were examined in the discipline of psychology as well as the field of visualization. Following a brief discussion on some latest developments in psychology, it outlines challenges and opportunities in making new discoveries about visualization through empirical studies.
We present STUDIES, a new speech corpus for developing a voice agent that can speak in a friendly manner. Humans naturally control their speech prosody to empathize with each other. By incorporating this "empathetic dialogue" behavior into a spoken dialogue system, we can develop a voice agent that can respond to a user more naturally. We designed the STUDIES corpus to include a speaker who speaks with empathy for the interlocutor's emotion explicitly. We describe our methodology to construct an empathetic dialogue speech corpus and report the analysis results of the STUDIES corpus. We conducted a text-to-speech experiment to initially investigate how we can develop more natural voice agent that can tune its speaking style corresponding to the interlocutor's emotion. The results show that the use of interlocutor's emotion label and conversational context embedding can produce speech with the same degree of naturalness as that synthesized by using the agent's emotion label. Our project page of the STUDIES corpus is http://sython.org/Corpus/STUDIES.
Painting captions are often dry and simplistic which motivates us to describe a painting creatively in the style of Shakespearean prose. This is a difficult problem, since there does not exist a large supervised dataset from paintings to Shakespearean prose. Our solution is to use an intermediate English poem description of the painting and then apply language style transfer which results in Shakespearean prose describing the painting. We rate our results by human evaluation on a Likert scale, and evaluate the quality of language style transfer using BLEU score as a function of prose length. We demonstrate the applicability and limitations of our approach by generating Shakespearean prose for famous paintings. We make our models and code publicly available.
Case study research has become an important research methodology for exploring phenomena in their natural contexts. Case studies have earned a distinct role in the empirical analysis of software engineering phenomena which are difficult to capture in isolation. Such phenomena often appear in the context of methods and development processes for which it is difficult to run large, controlled experiments as they usually have to reduce the scale in several respects and, hence, are detached from the reality of industrial software development. The other side of the medal is that the realistic socio-economic environments where we conduct case studies -- with real-life cases and realistic conditions -- also pose a plethora of practical challenges to planning and conducting case studies. In this experience report, we discuss such practical challenges and the lessons we learnt in conducting case studies in industry. Our goal is to help especially inexperienced researchers facing their first case studies in industry by increasing their awareness for typical obstacles they might face and practical ways to deal with those obstacles.