Search

Named entity recognition is an important task when constructing knowledge bases from unstructured data sources. Whereas entity detection methods mostly rely on extensive training data, Large Language Models (LLMs) have paved the way towards approaches that rely on zero-shot learning (ZSL) or few-shot learning (FSL) by taking advantage of the capabilities LLMs acquired during pretraining. Specifically, in very specialized scenarios where large-scale training data is not available, ZSL / FSL opens new opportunities. This paper follows this recent trend and investigates the potential of leveraging Large Language Models (LLMs) in such scenarios to automatically detect datasets and software within textual content from GitHub repositories. While existing methods focused solely on named entities, this study aims to broaden the scope by incorporating resources such as repositories and online hubs where entities are also represented by URLs. The study explores different FSL prompt learning approaches to enhance the LLMs' ability to identify dataset and software mentions within repository texts. Through analyses of LLM effectiveness and learning strategies, this paper offers insights into the potential of advanced language models for automated entity detection., Comment: accepted by KDD2024 workshop DL4KG

Multilingual text-to-speech systems convert text into speech across multiple languages. In many cases, text sentences may contain segments in different languages, a phenomenon known as code-switching. This is particularly common in Indonesia, especially between Indonesian and English. Despite its significance, no research has yet developed a multilingual TTS system capable of handling code-switching between these two languages. This study addresses Indonesian-English code-switching in STEN-TTS. Key modifications include adding a language identification component to the text-to-phoneme conversion using finetuned BERT for per-word language identification, as well as removing language embedding from the base model. Experimental results demonstrate that the code-switching model achieves superior naturalness and improved speech intelligibility compared to the Indonesian and English baseline STEN-TTS models., Comment: Accepted at O-COCOSDA 2024

Continual learning for automatic speech recognition (ASR) systems poses a challenge, especially with the need to avoid catastrophic forgetting while maintaining performance on previously learned tasks. This paper introduces a novel approach leveraging the machine speech chain framework to enable continual learning in ASR using gradient episodic memory (GEM). By incorporating a text-to-speech (TTS) component within the machine speech chain, we support the replay mechanism essential for GEM, allowing the ASR model to learn new tasks sequentially without significant performance degradation on earlier tasks. Our experiments, conducted on the LJ Speech dataset, demonstrate that our method outperforms traditional fine-tuning and multitask learning approaches, achieving a substantial error rate reduction while maintaining high performance across varying noise conditions. We showed the potential of our semi-supervised machine speech chain approach for effective and efficient continual learning in speech recognition., Comment: Published as a conference paper at O-COCOSDA 2024. 6 pages; 2 figures

An ideal speech recognition model has the capability to transcribe speech accurately under various characteristics of speech signals, such as speaking style (read and spontaneous), speech context (formal and informal), and background noise conditions (clean and moderate). Building such a model requires a significant amount of training data with diverse speech characteristics. Currently, Indonesian data is dominated by read, formal, and clean speech, leading to a scarcity of Indonesian data with other speech variabilities. To develop Indonesian automatic speech recognition (ASR), we present our research on state-of-the-art speech recognition models, namely Massively Multilingual Speech (MMS) and Whisper, as well as compiling a dataset comprising Indonesian speech with variabilities to facilitate our study. We further investigate the models' predictive ability to transcribe Indonesian speech data across different variability groups. The best results were achieved by the Whisper fine-tuned model across datasets with various characteristics, as indicated by the decrease in word error rate (WER) and character error rate (CER). Moreover, we found that speaking style variability affected model performance the most., Comment: Accepted at O-COCOSDA 2024

As Large Language Model (LLM)-based agents become increasingly autonomous and will more freely interact with each other, studying interactions between them becomes crucial to anticipate emergent phenomena and potential risks. Drawing inspiration from the widely popular Stanford Prison Experiment, we contribute to this line of research by studying interaction patterns of LLM agents in a context characterized by strict social hierarchy. We do so by specifically studying two types of phenomena: persuasion and anti-social behavior in simulated scenarios involving a guard and a prisoner agent who seeks to achieve a specific goal (i.e., obtaining additional yard time or escape from prison). Leveraging 200 experimental scenarios for a total of 2,000 machine-machine conversations across five different popular LLMs, we provide a set of noteworthy findings. We first document how some models consistently fail in carrying out a conversation in our multi-agent setup where power dynamics are at play. Then, for the models that were able to engage in successful interactions, we empirically show how the goal that an agent is set to achieve impacts primarily its persuasiveness, while having a negligible effect with respect to the agent's anti-social behavior. Third, we highlight how agents' personas, and particularly the guard's personality, drive both the likelihood of successful persuasion from the prisoner and the emergence of anti-social behaviors. Fourth, we show that even without explicitly prompting for specific personalities, anti-social behavior emerges by simply assigning agents' roles. These results bear implications for the development of interactive LLM agents as well as the debate on their societal impact.

Numerous methods and pipelines have recently emerged for the automatic extraction of knowledge graphs from documents such as scientific publications and patents. However, adapting these methods to incorporate alternative text sources like micro-blogging posts and news has proven challenging as they struggle to model open-domain entities and relations, typically found in these sources. In this paper, we propose an enhanced information extraction pipeline tailored to the extraction of a knowledge graph comprising open-domain entities from micro-blogging posts on social media platforms. Our pipeline leverages dependency parsing and classifies entity relations in an unsupervised manner through hierarchical clustering over word embeddings. We provide a use case on extracting semantic triples from a corpus of 100 thousand tweets about digital transformation and publicly release the generated knowledge graph. On the same dataset, we conduct two experimental evaluations, showing that the system produces triples with precision over 95% and outperforms similar pipelines of around 5% in terms of precision, while generating a comparatively higher number of triples., Comment: 42 pages, 6 figures

Hybrid models composing mechanistic ODE-based dynamics with flexible and expressive neural network components have grown rapidly in popularity, especially in scientific domains where such ODE-based modeling offers important interpretability and validated causal grounding (e.g., for counterfactual reasoning). The incorporation of mechanistic models also provides inductive bias in standard blackbox modeling approaches, critical when learning from small datasets or partially observed, complex systems. Unfortunately, as the hybrid models become more flexible, the causal grounding provided by the mechanistic model can quickly be lost. We address this problem by leveraging another common source of domain knowledge: \emph{ranking} of treatment effects for a set of interventions, even if the precise treatment effect is unknown. We encode this information in a \emph{causal loss} that we combine with the standard predictive loss to arrive at a \emph{hybrid loss} that biases our learning towards causally valid hybrid models. We demonstrate our ability to achieve a win-win, state-of-the-art predictive performance \emph{and} causal validity, in the challenging task of modeling glucose dynamics post-exercise in individuals with type 1 diabetes.

We parametrise the gauge-invariant ideals of the Toeplitz-Nica-Pimsner algebra of a strong compactly aligned product system over $\mathbb{Z}_+^d$ by using $2^d$-tuples of ideals of the coefficient algebra that are invariant, partially ordered, and maximal. We give an algebraic characterisation of maximality that allows the iteration of a $2^d$-tuple to the maximal one inducing the same gauge-invariant ideal. The parametrisation respects inclusions and intersections, while we characterise the join operation on the $2^d$-tuples that renders the parametrisation a lattice isomorphism. The problem of the parametrisation of the gauge-invariant ideals is equivalent to the study of relative Cuntz-Nica-Pimsner algebras, for which we provide a generalised Gauge-Invariant Uniqueness Theorem. We focus further on equivariant quotients of the Cuntz-Nica-Pimsner algebra and provide applications to regular product systems, C*-dynamical systems, strong finitely aligned higher-rank graphs, and product systems on finite frames. In particular, we provide a description of the parametrisation for (possibly non-automorphic) C*-dynamical systems and row-finite higher-rank graphs, which squares with known results when restricting to crossed products and to locally convex row-finite higher-rank graphs., Comment: 105 pages; editorial changes in the presentation of the paper and references added

The unprecedented success of image reconstruction approaches based on deep neural networks has revolutionised both the processing and the analysis paradigms in several applied disciplines. In the field of digital humanities, the task of digital reconstruction of ancient frescoes is particularly challenging due to the scarce amount of available training data caused by ageing, wear, tear and retouching over time. To overcome these difficulties, we consider the Deep Image Prior (DIP) inpainting approach which computes appropriate reconstructions by relying on the progressive updating of an untrained convolutional neural network so as to match the reliable piece of information in the image at hand while promoting regularisation elsewhere. In comparison with state-of-the-art approaches (based on variational/PDEs and patch-based methods), DIP-based inpainting reduces artefacts and better adapts to contextual/non-local information, thus providing a valuable and effective tool for art historians. As a case study, we apply such approach to reconstruct missing image contents in a dataset of highly damaged digital images of medieval paintings located into several chapels in the Mediterranean Alpine Arc and provide a detailed description on how visible and invisible (e.g., infrared) information can be integrated for identifying and reconstructing damaged image regions., Comment: 26 pages

Neural captioners are typically trained to mimic human-generated references without optimizing for any specific communication goal, leading to problems such as the generation of vague captions. In this paper, we show that fine-tuning an out-of-the-box neural captioner with a self-supervised discriminative communication objective helps to recover a plain, visually descriptive language that is more informative about image contents. Given a target image, the system must learn to produce a description that enables an out-of-the-box text-conditioned image retriever to identify such image among a set of candidates. We experiment with the popular ClipCap captioner, also replicating the main results with BLIP. In terms of similarity to ground-truth human descriptions, the captions emerging from discriminative finetuning lag slightly behind those generated by the non-finetuned model, when the latter is trained and tested on the same caption dataset. However, when the model is used without further tuning to generate captions for out-of-domain datasets, our discriminatively-finetuned captioner generates descriptions that resemble human references more than those produced by the same captioner without finetuning. We further show that, on the Conceptual Captions dataset, discriminatively finetuned captions are more helpful than either vanilla ClipCap captions or ground-truth captions for human annotators tasked with an image discrimination task., Comment: CVPR 2023

We study whether automatically-induced prompts that effectively extract information from a language model can also be used, out-of-the-box, to probe other language models for the same information. After confirming that discrete prompts induced with the AutoPrompt algorithm outperform manual and semi-manual prompts on the slot-filling task, we demonstrate a drop in performance for AutoPrompt prompts learned on a model and tested on another. We introduce a way to induce prompts by mixing language models at training time that results in prompts that generalize well across models. We conduct an extensive analysis of the induced prompts, finding that the more general prompts include a larger proportion of existing English words and have a less order-dependent and more uniform distribution of information across their component tokens. Our work provides preliminary evidence that it's possible to generate discrete prompts that can be induced once and used with a number of different models, and gives insights on the properties characterizing such prompts., Comment: Published as conference paper at ICLR 2023

This survey reviews works in which language models (LMs) are augmented with reasoning skills and the ability to use tools. The former is defined as decomposing a potentially complex task into simpler subtasks while the latter consists in calling external modules such as a code interpreter. LMs can leverage these augmentations separately or in combination via heuristics, or learn to do so from demonstrations. While adhering to a standard missing tokens prediction objective, such augmented LMs can use various, possibly non-parametric external modules to expand their context processing ability, thus departing from the pure language modeling paradigm. We therefore refer to them as Augmented Language Models (ALMs). The missing token objective allows ALMs to learn to reason, use tools, and even act, while still performing standard natural language tasks and even outperforming most regular LMs on several benchmarks. In this work, after reviewing current advance in ALMs, we conclude that this new research direction has the potential to address common limitations of traditional LMs such as interpretability, consistency, and scalability issues.

Language models (LMs) exhibit remarkable abilities to solve new tasks from just a few examples or textual instructions, especially at scale. They also, paradoxically, struggle with basic functionality, such as arithmetic or factual lookup, where much simpler and smaller models excel. In this paper, we show that LMs can teach themselves to use external tools via simple APIs and achieve the best of both worlds. We introduce Toolformer, a model trained to decide which APIs to call, when to call them, what arguments to pass, and how to best incorporate the results into future token prediction. This is done in a self-supervised way, requiring nothing more than a handful of demonstrations for each API. We incorporate a range of tools, including a calculator, a Q\&A system, two different search engines, a translation system, and a calendar. Toolformer achieves substantially improved zero-shot performance across a variety of downstream tasks, often competitive with much larger models, without sacrificing its core language modeling abilities.

As large pre-trained image-processing neural networks are being embedded in autonomous agents such as self-driving cars or robots, the question arises of how such systems can communicate with each other about the surrounding world, despite their different architectures and training regimes. As a first step in this direction, we systematically explore the task of referential communication in a community of heterogeneous state-of-the-art pre-trained visual networks, showing that they can develop, in a self-supervised way, a shared protocol to refer to a target object among a set of candidates. This shared protocol can also be used, to some extent, to communicate about previously unseen object categories of different granularity. Moreover, a visual network that was not initially part of an existing community can learn the community's protocol with remarkable ease. Finally, we study, both qualitatively and quantitatively, the properties of the emergent protocol, providing some evidence that it is capturing high-level semantic features of objects.

We compare the 0-shot performance of a neural caption-based image retriever when given as input either human-produced captions or captions generated by a neural captioner. We conduct this comparison on the recently introduced ImageCoDe data-set (Krojer et al., 2022) which contains hard distractors nearly identical to the images to be retrieved. We find that the neural retriever has much higher performance when fed neural rather than human captions, despite the fact that the former, unlike the latter, were generated without awareness of the distractors that make the task hard. Even more remarkably, when the same neural captions are given to human subjects, their retrieval performance is almost at chance level. Our results thus add to the growing body of evidence that, even when the ``language'' of neural models resembles English, this superficial resemblance might be deeply misleading., Comment: Accepted as a short paper at EMNLP 2022

Audio-based pornographic detection enables efficient adult content filtering without sacrificing performance by exploiting distinct spectral characteristics. To improve it, we explore pornographic sound modeling based on different neural architectures and acoustic features. We find that CNN trained on log mel spectrogram achieves the best performance on Pornography-800 dataset. Our experiment results also show that log mel spectrogram allows better representations for the models to recognize pornographic sounds. Finally, to classify whole audio waveforms rather than segments, we employ voting segment-to-audio technique that yields the best audio-level detection results., Comment: Published in AudioMostly 2022, ACM

Despite their practical success, modern seq2seq architectures are unable to generalize systematically on several SCAN tasks. Hence, it is not clear if SCAN-style compositional generalization is useful in realistic NLP tasks. In this work, we study the benefit that such compositionality brings about to several machine translation tasks. We present several focused modifications of Transformer that greatly improve generalization capabilities on SCAN and select one that remains on par with a vanilla Transformer on a standard machine translation (MT) task. Next, we study its performance in low-resource settings and on a newly introduced distribution-shifted English-French translation task. Overall, we find that improvements of a SCAN-capable model do not directly transfer to the resource-rich MT setup. In contrast, in the low-resource setup, general modifications lead to an improvement of up to 13.1% BLEU score w.r.t. a vanilla Transformer. Similarly, an improvement of 14% in an accuracy-based metric is achieved in the introduced compositional English-French translation task. This provides experimental evidence that the compositional generalization assessed in SCAN is particularly useful in resource-starved and domain-shifted scenarios., Comment: BlackboxNLP workshop, EMNLP 2021

The field of Materials Science is concerned with, e.g., properties and performance of materials. An important class of materials are crystalline materials that usually contain ``dislocations'' -- a line-like defect type. Dislocation decisively determine many important materials properties. Over the past decades, significant effort was put into understanding dislocation behavior across different length scales both with experimental characterization techniques as well as with simulations. However, for describing such dislocation structures there is still a lack of a common standard to represent and to connect dislocation domain knowledge across different but related communities. An ontology offers a common foundation to enable knowledge representation and data interoperability, which are important components to establish a ``digital twin''. This paper outlines the first steps towards the design of an ontology in the dislocation domain and shows a connection with the already existing ontologies in the materials science and engineering domain.

As deep networks begin to be deployed as autonomous agents, the issue of how they can communicate with each other becomes important. Here, we train two deep nets from scratch to perform realistic referent identification through unsupervised emergent communication. We show that the largely interpretable emergent protocol allows the nets to successfully communicate even about object types they did not see at training time. The visual representations induced as a by-product of our training regime, moreover, show comparable quality, when re-used as generic visual features, to a recent self-supervised learning model. Our results provide concrete evidence of the viability of (interpretable) emergent deep net communication in a more realistic scenario than previously considered, as well as establishing an intriguing link between this field and self-supervised visual learning., Comment: Accepted at NeurIPS 2021

Today, we are seeing an ever-increasing number of clinical notes that contain clinical results, images, and textual descriptions of patient's health state. All these data can be analyzed and employed to cater novel services that can help people and domain experts with their common healthcare tasks. However, many technologies such as Deep Learning and tools like Word Embeddings have started to be investigated only recently, and many challenges remain open when it comes to healthcare domain applications. To address these challenges, we propose the use of Deep Learning and Word Embeddings for identifying sixteen morbidity types within textual descriptions of clinical records. For this purpose, we have used a Deep Learning model based on Bidirectional Long-Short Term Memory (LSTM) layers which can exploit state-of-the-art vector representations of data such as Word Embeddings. We have employed pre-trained Word Embeddings namely GloVe and Word2Vec, and our own Word Embeddings trained on the target domain. Furthermore, we have compared the performances of the deep learning approaches against the traditional tf-idf using Support Vector Machine and Multilayer perceptron (our baselines). From the obtained results it seems that the latter outperforms the combination of Deep Learning approaches using any word embeddings. Our preliminary results indicate that there are specific features that make the dataset biased in favour of traditional machine learning approaches., Comment: 12 pages, 2 figures, 2 tables, SmartPhil 2020-First Workshop on Smart Personal Health Interfaces, Associated to ACM IUI 2020

The continuous growth of scientific literature brings innovations and, at the same time, raises new challenges. One of them is related to the fact that its analysis has become difficult due to the high volume of published papers for which manual effort for annotations and management is required. Novel technological infrastructures are needed to help researchers, research policy makers, and companies to time-efficiently browse, analyse, and forecast scientific research. Knowledge graphs i.e., large networks of entities and relationships, have proved to be effective solution in this space. Scientific knowledge graphs focus on the scholarly domain and typically contain metadata describing research publications such as authors, venues, organizations, research topics, and citations. However, the current generation of knowledge graphs lacks of an explicit representation of the knowledge presented in the research papers. As such, in this paper, we present a new architecture that takes advantage of Natural Language Processing and Machine Learning methods for extracting entities and relationships from research publications and integrates them in a large-scale knowledge graph. Within this research work, we i) tackle the challenge of knowledge extraction by employing several state-of-the-art Natural Language Processing and Text Mining tools, ii) describe an approach for integrating entities and relationships generated by these tools, iii) show the advantage of such an hybrid system over alternative approaches, and vi) as a chosen use case, we generated a scientific knowledge graph including 109,105 triples, extracted from 26,827 abstracts of papers within the Semantic Web domain. As our approach is general and can be applied to any domain, we expect that it can facilitate the management, analysis, dissemination, and processing of scientific knowledge., Comment: Accepted for publication in Future Generation Computer Systems journal - Special Issue on Machine Learning and Knowledge Graphs

Molecular-motor generated active stresses drive the cytoskeleton away from equilibrium, endowing it with tunable mechanical properties that are essential for diverse functions such as cell division and motility[1-5]. Designing analogous biomimetic systems is a key prerequisite for creating active matter that can emulate cellular functions[6-7]. These long-term goals requires understanding of how motor-generated stresses tune the mechanics of filamentous networks[8-11]. In microtubule-based active matter, kinesin motors generate extensile motion that leads to persistent breaking and reforming of the network links[12]. We study how such microscopic dynamics modifies the network's mechanical properties, uncovering that the network viscosity first increases with the imposed shear rate before transitioning back to a low-viscosity state. The non-monotonic shear-dependent viscosity can be controlled by tuning the speed of molecular motors. A two-state phenomenological model that incorporates liquid- and solid-like elements quantitatively relates the non-monotonic shear-rate-dependent viscosity to locally-measured flows. These studies show that rheology of extensile networks are different from previously studied active gels[13], where contractility enhances mechanical stiffness. Moreover, the flow induced gelation is not captured by continuum models of hydrodynamically interacting swimmers[14-21]. Observation of activity-dependent viscoelasticity necessitates the development of models for self-yielding of soft active solids whose intrinsic active stresses fluidize or stiffen the network.

Research in multi-agent cooperation has shown that artificial agents are able to learn to play a simple referential game while developing a shared lexicon. This lexicon is not easy to analyze, as it does not show many properties of a natural language. In a simple referential game with two neural network-based agents, we analyze the object-symbol mapping trying to understand what kind of strategy was used to develop the emergent language. We see that, when the environment is uniformly distributed, the agents rely on a random subset of features to describe the objects. When we modify the objects making one feature non-uniformly distributed,the agents realize it is less informative and start to ignore it, and, surprisingly, they make a better use of the remaining features. This interesting result suggests that more natural, less uniformly distributed environments might aid in spurring the emergence of better-behaved languages., Comment: 3rd NeurIPS Workshop on Emergent Communication

Lake and Baroni (2018) introduced the SCAN dataset probing the ability of seq2seq models to capture compositional generalizations, such as inferring the meaning of "jump around" 0-shot from the component words. Recurrent networks (RNNs) were found to completely fail the most challenging generalization cases. We test here a convolutional network (CNN) on these tasks, reporting hugely improved performance with respect to RNNs. Despite the big improvement, the CNN has however not induced systematic rules, suggesting that the difference between compositional and non-compositional behaviour is not clear-cut., Comment: accepted as a short paper at ACL 2019

This paper describes FBK's submission to the end-to-end English-German speech translation task at IWSLT 2018. Our system relies on a state-of-the-art model based on LSTMs and CNNs, where the CNNs are used to reduce the temporal dimension of the audio input, which is in general much higher than machine translation input. Our model was trained only on the audio-to-text parallel data released for the task, and fine-tuned on cleaned subsets of the original training corpus. The addition of weight normalization and label smoothing improved the baseline system by 1.0 BLEU point on our validation set. The final submission also featured checkpoint averaging within a training run and ensemble decoding of models trained during multiple runs. On test data, our best single model obtained a BLEU score of 9.7, while the ensemble obtained a BLEU score of 10.24., Comment: 6 pages, 2 figures, system description at the 15th International Workshop on Spoken Language Translation (IWSLT) 2018

Reconstructing the 3D shape of an object from a set of images is a classical problem in Computer Vision. Photometric stereo is one of the possible approaches. It stands on the assumption that the object is observed from a fixed point of view under different lighting conditions. The traditional approach requires that the position of the light sources is accurately known. It has been proved that the lights position can be estimated directly from the data, when at least 6 images of the observed object are available. In this paper, we give a Matlab implementation of the algorithm for solving the photometric stereo problem under unknown lighting, and propose a simple shooting technique to solve the bas-relief ambiguity., Comment: new version