The robustness of multimodal deep learning models to realistic changes in the input text is critical for their applicability to important tasks such as text-to-image retrieval and cross-modal entailment. To measure robustness, several existing approaches edit the text data, but do so without leveraging the cross-modal information present in multimodal data. Information from the visual modality, such as color, size, and shape, provide additional attributes that users can include in their inputs. Thus, we propose cross-modal attribute insertions as a realistic perturbation strategy for vision-and-language data that inserts visual attributes of the objects in the image into the corresponding text (e.g., "girl on a chair" to "little girl on a wooden chair"). Our proposed approach for cross-modal attribute insertions is modular, controllable, and task-agnostic. We find that augmenting input text using cross-modal insertions causes state-of-the-art approaches for text-to-image retrieval and cross-modal entailment to perform poorly, resulting in relative drops of 15% in MRR and 20% in $F_1$ score, respectively. Crowd-sourced annotations demonstrate that cross-modal insertions lead to higher quality augmentations for multimodal data than augmentations using text-only data, and are equivalent in quality to original examples. We release the code to encourage robustness evaluations of deep vision-and-language models: //github.com/claws-lab/multimodal-robustness-xmai.
相關內容
Virtual try-on is a critical image synthesis task that aims to transfer clothes from one image to another while preserving the details of both humans and clothes. While many existing methods rely on Generative Adversarial Networks (GANs) to achieve this, flaws can still occur, particularly at high resolutions. Recently, the diffusion model has emerged as a promising alternative for generating high-quality images in various applications. However, simply using clothes as a condition for guiding the diffusion model to inpaint is insufficient to maintain the details of the clothes. To overcome this challenge, we propose an exemplar-based inpainting approach that leverages a warping module to guide the diffusion model's generation effectively. The warping module performs initial processing on the clothes, which helps to preserve the local details of the clothes. We then combine the warped clothes with clothes-agnostic person image and add noise as the input of diffusion model. Additionally, the warped clothes is used as local conditions for each denoising process to ensure that the resulting output retains as much detail as possible. Our approach, namely Diffusion-based Conditional Inpainting for Virtual Try-ON (DCI-VTON), effectively utilizes the power of the diffusion model, and the incorporation of the warping module helps to produce high-quality and realistic virtual try-on results. Experimental results on VITON-HD demonstrate the effectiveness and superiority of our method.
Critical points mark locations in the domain where the level-set topology of a scalar function undergoes fundamental changes and thus indicate potentially interesting features in the data. Established methods exist to locate and relate such points in a deterministic setting, but it is less well understood how the concept of critical points can be extended to the analysis of uncertain data. Most methods for this task aim at finding likely locations of critical points or estimate the probability of their occurrence locally but do not indicate if critical points at potentially different locations in different realizations of a stochastic process are manifestations of the same feature, which is required to characterize the spatial uncertainty of critical points. Previous work on relating critical points across different realizations reported challenges for interpreting the resulting spatial distribution of critical points but did not investigate the causes. In this work, we provide a mathematical formulation of the problem of finding critical points with spatial uncertainty and computing their spatial distribution, which leads us to the notion of uncertain critical points. We analyze the theoretical properties of these structures and highlight connections to existing works for special classes of uncertain fields. We derive conditions under which well-interpretable results can be obtained and discuss the implications of those restrictions for the field of visualization. We demonstrate that the discussed limitations are not purely academic but also arise in real-world data.
The ability to interpret machine learning models has become increasingly important as their usage in data science continues to rise. Most current interpretability methods are optimized to work on either (\textit{i}) a global scale, where the goal is to rank features based on their contributions to overall variation in an observed population, or (\textit{ii}) the local level, which aims to detail on how important a feature is to a particular individual in the data set. In this work, a new operator is proposed called the "GlObal And Local Score" (GOALS): a simple \textit{post hoc} approach to simultaneously assess local and global feature variable importance in nonlinear models. Motivated by problems in biomedicine, the approach is demonstrated using Gaussian process regression where the task of understanding how genetic markers are associated with disease progression both within individuals and across populations is of high interest. Detailed simulations and real data analyses illustrate the flexible and efficient utility of GOALS over state-of-the-art variable importance strategies.
Our modern world relies on a growing number of interconnected and interacting devices, leading to a plethora of logs establishing audit trails for all kinds of events. Simultaneously, logs become increasingly important for forensic investigations, and thus, an adversary will aim to alter logs to avoid culpability, e.g., by compromising devices that generate and store logs. Thus, it is essential to ensure that no one can tamper with any logs without going undetected. However, existing approaches to establish tamper evidence of logs do not scale and cannot protect the increasingly large number of devices found today, as they impose large storage or network overheads. Additionally, most schemes do not provide an efficient mechanism to prove that individual events have been logged to establish accountability when different devices interact. This paper introduces a novel scheme for practical large-scale tamper-evident logging with the help of a trusted third party. To achieve this, we present a new binary hash tree construction designed around timestamps to achieve constant storage overhead with a configured temporal resolution. Additionally, our design enables the efficient construction of shareable proofs, proving that an event was indeed logged. Our evaluation shows that - using practical parameters - our scheme can localize any tampering of logs with a sub-second resolution, with a constant overhead of ~8KB per hour per device.
In the growing domain of scientific machine learning, in-context operator learning has demonstrated notable potential in learning operators from prompted data during inference stage without weight updates. However, the current model's overdependence on sensor data, may inadvertently overlook the invaluable human insight into the operator. To address this, we present a transformation of in-context operator learning into a multi-modal paradigm. We propose the use of "captions" to integrate human knowledge about the operator, expressed through natural language descriptions and equations. We illustrate how this method not only broadens the flexibility and generality of physics-informed learning, but also significantly boosts learning performance and reduces data needs. Furthermore, we introduce a more efficient neural network architecture for multi-modal in-context operator learning, referred to as "ICON-LM", based on a language-model-like architecture. We demonstrate the viability of "ICON-LM" for scientific machine learning tasks, which creates a new path for the application of language models.
Data augmentation, the artificial creation of training data for machine learning by transformations, is a widely studied research field across machine learning disciplines. While it is useful for increasing the generalization capabilities of a model, it can also address many other challenges and problems, from overcoming a limited amount of training data over regularizing the objective to limiting the amount data used to protect privacy. Based on a precise description of the goals and applications of data augmentation (C1) and a taxonomy for existing works (C2), this survey is concerned with data augmentation methods for textual classification and aims to achieve a concise and comprehensive overview for researchers and practitioners (C3). Derived from the taxonomy, we divided more than 100 methods into 12 different groupings and provide state-of-the-art references expounding which methods are highly promising (C4). Finally, research perspectives that may constitute a building block for future work are given (C5).
The inductive biases of graph representation learning algorithms are often encoded in the background geometry of their embedding space. In this paper, we show that general directed graphs can be effectively represented by an embedding model that combines three components: a pseudo-Riemannian metric structure, a non-trivial global topology, and a unique likelihood function that explicitly incorporates a preferred direction in embedding space. We demonstrate the representational capabilities of this method by applying it to the task of link prediction on a series of synthetic and real directed graphs from natural language applications and biology. In particular, we show that low-dimensional cylindrical Minkowski and anti-de Sitter spacetimes can produce equal or better graph representations than curved Riemannian manifolds of higher dimensions.
Influenced by the stunning success of deep learning in computer vision and language understanding, research in recommendation has shifted to inventing new recommender models based on neural networks. In recent years, we have witnessed significant progress in developing neural recommender models, which generalize and surpass traditional recommender models owing to the strong representation power of neural networks. In this survey paper, we conduct a systematic review on neural recommender models, aiming to summarize the field to facilitate future progress. Distinct from existing surveys that categorize existing methods based on the taxonomy of deep learning techniques, we instead summarize the field from the perspective of recommendation modeling, which could be more instructive to researchers and practitioners working on recommender systems. Specifically, we divide the work into three types based on the data they used for recommendation modeling: 1) collaborative filtering models, which leverage the key source of user-item interaction data; 2) content enriched models, which additionally utilize the side information associated with users and items, like user profile and item knowledge graph; and 3) context enriched models, which account for the contextual information associated with an interaction, such as time, location, and the past interactions. After reviewing representative works for each type, we finally discuss some promising directions in this field, including benchmarking recommender systems, graph reasoning based recommendation models, and explainable and fair recommendations for social good.
The canonical approach to video-and-language learning (e.g., video question answering) dictates a neural model to learn from offline-extracted dense video features from vision models and text features from language models. These feature extractors are trained independently and usually on tasks different from the target domains, rendering these fixed features sub-optimal for downstream tasks. Moreover, due to the high computational overload of dense video features, it is often difficult (or infeasible) to plug feature extractors directly into existing approaches for easy finetuning. To provide a remedy to this dilemma, we propose a generic framework ClipBERT that enables affordable end-to-end learning for video-and-language tasks, by employing sparse sampling, where only a single or a few sparsely sampled short clips from a video are used at each training step. Experiments on text-to-video retrieval and video question answering on six datasets demonstrate that ClipBERT outperforms (or is on par with) existing methods that exploit full-length videos, suggesting that end-to-end learning with just a few sparsely sampled clips is often more accurate than using densely extracted offline features from full-length videos, proving the proverbial less-is-more principle. Videos in the datasets are from considerably different domains and lengths, ranging from 3-second generic domain GIF videos to 180-second YouTube human activity videos, showing the generalization ability of our approach. Comprehensive ablation studies and thorough analyses are provided to dissect what factors lead to this success. Our code is publicly available at //github.com/jayleicn/ClipBERT
Due to the significance and value in human-computer interaction and natural language processing, task-oriented dialog systems are attracting more and more attention in both academic and industrial communities. In this paper, we survey recent advances and challenges in an issue-specific manner. We discuss three critical topics for task-oriented dialog systems: (1) improving data efficiency to facilitate dialog system modeling in low-resource settings, (2) modeling multi-turn dynamics for dialog policy learning to achieve better task-completion performance, and (3) integrating domain ontology knowledge into the dialog model in both pipeline and end-to-end models. We also review the recent progresses in dialog evaluation and some widely-used corpora. We believe that this survey can shed a light on future research in task-oriented dialog systems.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191