In this study we approach the complexity of the vaccine debate from a new and comprehensive perspective. Focusing on the Italian context, we examine almost all the online information produced in the 2016-2021 timeframe by both sources that have a reputation for misinformation and those that do not. Although reliable sources can rely on larger newsrooms and cover more news than misinformation ones, the transfer entropy analysis of the corresponding time series reveals that the former have not always informationally dominated the latter on the vaccine subject. Indeed, the pre-pandemic period sees misinformation establish itself as leader of the process, even in causal terms, and gain dramatically more user engagement than news from reliable sources. Despite this information gap was filled during the Covid-19 outbreak, the newfound leading role of reliable sources as drivers of the information ecosystem has only partially had a beneficial effect in reducing user engagement with misinformation on vaccines. Our results indeed show that, except for effectiveness of vaccination, reliable sources have never adequately countered the anti-vax narrative, specially in the pre-pandemic period, thus contributing to exacerbate science denial and belief in conspiracy theories. At the same time, however, they confirm the efficacy of assiduously proposing a convincing counter-narrative to misinformation spread. Indeed, effectiveness of vaccination turns out to be the least engaging topic discussed by misinformation during the pandemic period, when compared to other polarising arguments such as safety concerns, legal issues and vaccine business. By highlighting the strengths and weaknesses of institutional and mainstream communication, our findings can be a valuable asset for improving and better targeting campaigns against misinformation on vaccines.
Realizing the potential gains of large-scale MIMO systems requires the accurate estimation of their channels or the fine adjustment of their narrow beams. This, however, is typically associated with high channel acquisition/beam sweeping overhead that scales with the number of antennas. Machine and deep learning represent promising approaches to overcome these challenges thanks to their powerful ability to learn from prior observations and side information. Training machine and deep learning models, however, requires large-scale datasets that are expensive to collect in deployed systems. To address this challenge, we propose a novel direction that utilizes digital replicas of the physical world to reduce or even eliminate the MIMO channel acquisition overhead. In the proposed digital twin aided communication, 3D models that approximate the real-world communication environment are constructed and accurate ray-tracing is utilized to simulate the site-specific channels. These channels can then be used to aid various communication tasks. Further, we propose to use machine learning to approximate the digital replicas and reduce the ray tracing computational cost. To evaluate the proposed digital twin based approach, we conduct a case study focusing on the position-aided beam prediction task. The results show that a learning model trained solely with the data generated by the digital replica can achieve relatively good performance on the real-world data. Moreover, a small number of real-world data points can quickly achieve near-optimal performance, overcoming the modeling mismatches between the physical and digital worlds and significantly reducing the data acquisition overhead.
The task of text-to-SQL parsing, which aims at converting natural language questions into executable SQL queries, has garnered increasing attention in recent years, as it can assist end users in efficiently extracting vital information from databases without the need for technical background. One of the major challenges in text-to-SQL parsing is domain generalization, i.e., how to generalize well to unseen databases. Recently, the pre-trained text-to-text transformer model, namely T5, though not specialized for text-to-SQL parsing, has achieved state-of-the-art performance on standard benchmarks targeting domain generalization. In this work, we explore ways to further augment the pre-trained T5 model with specialized components for text-to-SQL parsing. Such components are expected to introduce structural inductive bias into text-to-SQL parsers thus improving model's capacity on (potentially multi-hop) reasoning, which is critical for generating structure-rich SQLs. To this end, we propose a new architecture GRAPHIX-T5, a mixed model with the standard pre-trained transformer model augmented by some specially-designed graph-aware layers. Extensive experiments and analysis demonstrate the effectiveness of GRAPHIX-T5 across four text-to-SQL benchmarks: SPIDER, SYN, REALISTIC and DK. GRAPHIX-T5 surpass all other T5-based parsers with a significant margin, achieving new state-of-the-art performance. Notably, GRAPHIX-T5-large reach performance superior to the original T5-large by 5.7% on exact match (EM) accuracy and 6.6% on execution accuracy (EX). This even outperforms the T5-3B by 1.2% on EM and 1.5% on EX.
AI explanations are often mentioned as a way to improve human-AI decision-making. Yet, empirical studies have not found consistent evidence of explanations' effectiveness and, on the contrary, suggest that they can increase overreliance when the AI system is wrong. While many factors may affect reliance on AI support, one important factor is how decision-makers reconcile their own intuition -- which may be based on domain knowledge, prior task experience, or pattern recognition -- with the information provided by the AI system to determine when to override AI predictions. We conduct a think-aloud, mixed-methods study with two explanation types (feature- and example-based) for two prediction tasks to explore how decision-makers' intuition affects their use of AI predictions and explanations, and ultimately their choice of when to rely on AI. Our results identify three types of intuition involved in reasoning about AI predictions and explanations: intuition about the task outcome, features, and AI limitations. Building on these, we summarize three observed pathways for decision-makers to apply their own intuition and override AI predictions. We use these pathways to explain why (1) the feature-based explanations we used did not improve participants' decision outcomes and increased their overreliance on AI, and (2) the example-based explanations we used improved decision-makers' performance over feature-based explanations and helped achieve complementary human-AI performance. Overall, our work identifies directions for further development of AI decision-support systems and explanation methods that help decision-makers effectively apply their intuition to achieve appropriate reliance on AI.
Visual objects often have acoustic signatures that are naturally synchronized with them in audio-bearing video recordings. For this project, we explore the multimodal feature aggregation for video instance segmentation task, in which we integrate audio features into our video segmentation model to conduct an audio-visual learning scheme. Our method is based on existing video instance segmentation method which leverages rich contextual information across video frames. Since this is the first attempt to investigate the audio-visual instance segmentation, a novel dataset, including 20 vocal classes with synchronized video and audio recordings, is collected. By utilizing combined decoder to fuse both video and audio features, our model shows a slight improvements compared to the base model. Additionally, we managed to show the effectiveness of different modules by conducting extensive ablations.
To study the behavior of freely moving model organisms such as zebrafish (Danio rerio) and fruit flies (Drosophila) across multiple spatial scales, it would be ideal to use a light microscope that can resolve 3D information over a wide field of view (FOV) at high speed and high spatial resolution. However, it is challenging to design an optical instrument to achieve all of these properties simultaneously. Existing techniques for large-FOV microscopic imaging and for 3D image measurement typically require many sequential image snapshots, thus compromising speed and throughput. Here, we present 3D-RAPID, a computational microscope based on a synchronized array of 54 cameras that can capture high-speed 3D topographic videos over a 135-cm^2 area, achieving up to 230 frames per second at throughputs exceeding 5 gigapixels (GPs) per second. 3D-RAPID features a 3D reconstruction algorithm that, for each synchronized temporal snapshot, simultaneously fuses all 54 images seamlessly into a globally-consistent composite that includes a coregistered 3D height map. The self-supervised 3D reconstruction algorithm itself trains a spatiotemporally-compressed convolutional neural network (CNN) that maps raw photometric images to 3D topography, using stereo overlap redundancy and ray-propagation physics as the only supervision mechanism. As a result, our end-to-end 3D reconstruction algorithm is robust to generalization errors and scales to arbitrarily long videos from arbitrarily sized camera arrays. The scalable hardware and software design of 3D-RAPID addresses a longstanding problem in the field of behavioral imaging, enabling parallelized 3D observation of large collections of freely moving organisms at high spatiotemporal throughputs, which we demonstrate in ants (Pogonomyrmex barbatus), fruit flies, and zebrafish larvae.
Current interpretability methods focus on explaining a particular model's decision through present input features. Such methods do not inform the user of the sufficient conditions that alter these decisions when they are not desirable. Contrastive explanations circumvent this problem by providing explanations of the form "If the feature $X>x$, the output $Y$ would be different''. While different approaches are developed to find contrasts; these methods do not all deal with mutability and attainability constraints. In this work, we present a novel approach to locally contrast the prediction of any classifier. Our Contrastive Entropy-based explanation method, CEnt, approximates a model locally by a decision tree to compute entropy information of different feature splits. A graph, G, is then built where contrast nodes are found through a one-to-many shortest path search. Contrastive examples are generated from the shortest path to reflect feature splits that alter model decisions while maintaining lower entropy. We perform local sampling on manifold-like distances computed by variational auto-encoders to reflect data density. CEnt is the first non-gradient-based contrastive method generating diverse counterfactuals that do not necessarily exist in the training data while satisfying immutability (ex. race) and semi-immutability (ex. age can only change in an increasing direction). Empirical evaluation on four real-world numerical datasets demonstrates the ability of CEnt in generating counterfactuals that achieve better proximity rates than existing methods without compromising latency, feasibility, and attainability. We further extend CEnt to imagery data to derive visually appealing and useful contrasts between class labels on MNIST and Fashion MNIST datasets. Finally, we show how CEnt can serve as a tool to detect vulnerabilities of textual classifiers.
Pre-training and fine-tuning have achieved significant advances in the information retrieval (IR). A typical approach is to fine-tune all the parameters of large-scale pre-trained models (PTMs) on downstream tasks. As the model size and the number of tasks increase greatly, such approach becomes less feasible and prohibitively expensive. Recently, a variety of parameter-efficient tuning methods have been proposed in natural language processing (NLP) that only fine-tune a small number of parameters while still attaining strong performance. Yet there has been little effort to explore parameter-efficient tuning for IR. In this work, we first conduct a comprehensive study of existing parameter-efficient tuning methods at both the retrieval and re-ranking stages. Unlike the promising results in NLP, we find that these methods cannot achieve comparable performance to full fine-tuning at both stages when updating less than 1\% of the original model parameters. More importantly, we find that the existing methods are just parameter-efficient, but not learning-efficient as they suffer from unstable training and slow convergence. To analyze the underlying reason, we conduct a theoretical analysis and show that the separation of the inserted trainable modules makes the optimization difficult. To alleviate this issue, we propose to inject additional modules alongside the \acp{PTM} to make the original scattered modules connected. In this way, all the trainable modules can form a pathway to smooth the loss surface and thus help stabilize the training process. Experiments at both retrieval and re-ranking stages show that our method outperforms existing parameter-efficient methods significantly, and achieves comparable or even better performance over full fine-tuning.
The recent trend for multi-camera 3D object detection is through the unified bird's-eye view (BEV) representation. However, directly transforming features extracted from the image-plane view to BEV inevitably results in feature distortion, especially around the objects of interest, making the objects blur into the background. To this end, we propose OA-BEV, a network that can be plugged into the BEV-based 3D object detection framework to bring out the objects by incorporating object-aware pseudo-3D features and depth features. Such features contain information about the object's position and 3D structures. First, we explicitly guide the network to learn the depth distribution by object-level supervision from each 3D object's center. Then, we select the foreground pixels by a 2D object detector and project them into 3D space for pseudo-voxel feature encoding. Finally, the object-aware depth features and pseudo-voxel features are incorporated into the BEV representation with a deformable attention mechanism. We conduct extensive experiments on the nuScenes dataset to validate the merits of our proposed OA-BEV. Our method achieves consistent improvements over the BEV-based baselines in terms of both average precision and nuScenes detection score. Our codes will be published.
The recently proposed Vision transformers (ViTs) have shown very impressive empirical performance in various computer vision tasks, and they are viewed as an important type of foundation model. However, ViTs are typically constructed with large-scale sizes, which then severely hinder their potential deployment in many practical resources-constrained applications. To mitigate this challenging problem, structured pruning is a promising solution to compress model size and enable practical efficiency. However, unlike its current popularity for CNNs and RNNs, structured pruning for ViT models is little explored. In this paper, we propose GOHSP, a unified framework of Graph and Optimization-based Structured Pruning for ViT models. We first develop a graph-based ranking for measuring the importance of attention heads, and the extracted importance information is further integrated to an optimization-based procedure to impose the heterogeneous structured sparsity patterns on the ViT models. Experimental results show that our proposed GOHSP demonstrates excellent compression performance. On CIFAR-10 dataset, our approach can bring 40% parameters reduction with no accuracy loss for ViT-Small model. On ImageNet dataset, with 30% and 35% sparsity ratio for DeiT-Tiny and DeiT-Small models, our approach achieves 1.65% and 0.76% accuracy increase over the existing structured pruning methods, respectively.