Explainable artificial intelligence (XAI) methods lack ground truth. In its place, method developers have relied on axioms to determine desirable properties for their explanations' behavior. For high stakes uses of machine learning that require explainability, it is not sufficient to rely on axioms as the implementation, or its usage, can fail to live up to the ideal. As a result, there exists active research on validating the performance of XAI methods. The need for validation is especially magnified in domains with a reliance on XAI. A procedure frequently used to assess their utility, and to some extent their fidelity, is an ablation study. By perturbing the input variables in rank order of importance, the goal is to assess the sensitivity of the model's performance. Perturbing important variables should correlate with larger decreases in measures of model capability than perturbing less important features. While the intent is clear, the actual implementation details have not been studied rigorously for tabular data. Using five datasets, three XAI methods, four baselines, and three perturbations, we aim to show 1) how varying perturbations and adding simple guardrails can help to avoid potentially flawed conclusions, 2) how treatment of categorical variables is an important consideration in both post-hoc explainability and ablation studies, and 3) how to identify useful baselines for XAI methods and viable perturbations for ablation studies.

Contextualized word embeddings can lead to state-of-the-art performances in natural language understanding. Recently, a pre-trained deep contextualized text encoder such as BERT has shown its potential in improving natural language tasks including abstractive summarization. Existing approaches in dialogue summarization focus on incorporating a large language model into summarization task trained on large-scale corpora consisting of news articles rather than dialogues of multiple speakers. In this paper, we introduce self-supervised methods to compensate shortcomings to train a dialogue summarization model. Our principle is to detect incoherent information flows using pretext dialogue text to enhance BERT's ability to contextualize the dialogue text representations. We build and fine-tune an abstractive dialogue summarization model on a shared encoder-decoder architecture using the enhanced BERT. We empirically evaluate our abstractive dialogue summarizer with the SAMSum corpus, a recently introduced dataset with abstractive dialogue summaries. All of our methods have contributed improvements to abstractive summary measured in ROUGE scores. Through an extensive ablation study, we also present a sensitivity analysis to critical model hyperparameters, probabilities of switching utterances and masking interlocutors.

In this work, we address the problem of generating speech from silent lip videos for any speaker in the wild. In stark contrast to previous works, our method (i) is not restricted to a fixed number of speakers, (ii) does not explicitly impose constraints on the domain or the vocabulary and (iii) deals with videos that are recorded in the wild as opposed to within laboratory settings. The task presents a host of challenges, with the key one being that many features of the desired target speech, like voice, pitch and linguistic content, cannot be entirely inferred from the silent face video. In order to handle these stochastic variations, we propose a new VAE-GAN architecture that learns to associate the lip and speech sequences amidst the variations. With the help of multiple powerful discriminators that guide the training process, our generator learns to synthesize speech sequences in any voice for the lip movements of any person. Extensive experiments on multiple datasets show that we outperform all baselines by a large margin. Further, our network can be fine-tuned on videos of specific identities to achieve a performance comparable to single-speaker models that are trained on $4\times$ more data. We conduct numerous ablation studies to analyze the effect of different modules of our architecture. We also provide a demo video that demonstrates several qualitative results along with the code and trained models on our website: \url{http://cvit.iiit.ac.in/research/projects/cvit-projects/lip-to-speech-synthesis}}

The processing of legal texts has been developing as an emerging field in natural language processing (NLP). Legal texts contain unique jargon and complex linguistic attributes in vocabulary, semantics, syntax, and morphology. Therefore, the development of text simplification (TS) methods specific to the legal domain is of paramount importance for facilitating comprehension of legal text by ordinary people and providing inputs to high-level models for mainstream legal NLP applications. While a recent study proposed a rule-based TS method for legal text, learning-based TS in the legal domain has not been considered previously. Here we introduce an unsupervised simplification method for legal texts (USLT). USLT performs domain-specific TS by replacing complex words and splitting long sentences. To this end, USLT detects complex words in a sentence, generates candidates via a masked-transformer model, and selects a candidate for substitution based on a rank score. Afterward, USLT recursively decomposes long sentences into a hierarchy of shorter core and context sentences while preserving semantic meaning. We demonstrate that USLT outperforms state-of-the-art domain-general TS methods in text simplicity while keeping the semantics intact.

An outstanding image-text retrieval model depends on high-quality labeled data. While the builders of existing image-text retrieval datasets strive to ensure that the caption matches the linked image, they cannot prevent a caption from fitting other images. We observe that such a many-to-many matching phenomenon is quite common in the widely-used retrieval datasets, where one caption can describe up to 178 images. These large matching-lost data not only confuse the model in training but also weaken the evaluation accuracy. Inspired by visual and textual entailment tasks, we propose a multi-modal entailment classifier to determine whether a sentence is entailed by an image plus its linked captions. Subsequently, we revise the image-text retrieval datasets by adding these entailed captions as additional weak labels of an image and develop a universal variable learning rate strategy to teach a retrieval model to distinguish the entailed captions from other negative samples. In experiments, we manually annotate an entailment-corrected image-text retrieval dataset for evaluation. The results demonstrate that the proposed entailment classifier achieves about 78% accuracy and consistently improves the performance of image-text retrieval baselines.

Large-scale graphs are ubiquitous in real-world scenarios and can be trained by Graph Neural Networks (GNNs) to generate representation for downstream tasks. Given the abundant information and complex topology of a large-scale graph, we argue that redundancy exists in such graphs and will degrade the training efficiency. Unfortunately, the model scalability severely restricts the efficiency of training large-scale graphs via vanilla GNNs. Despite recent advances in sampling-based training methods, sampling-based GNNs generally overlook the redundancy issue. It still takes intolerable time to train these models on large-scale graphs. Thereby, we propose to drop redundancy and improve efficiency of training large-scale graphs with GNNs, by rethinking the inherent characteristics in a graph. In this paper, we pioneer to propose a once-for-all method, termed DropReef, to drop the redundancy in large-scale graphs. Specifically, we first conduct preliminary experiments to explore potential redundancy in large-scale graphs. Next, we present a metric to quantify the neighbor heterophily of all nodes in a graph. Based on both experimental and theoretical analysis, we reveal the redundancy in a large-scale graph, i.e., nodes with high neighbor heterophily and a great number of neighbors. Then, we propose DropReef to detect and drop the redundancy in large-scale graphs once and for all, helping reduce the training time while ensuring no sacrifice in the model accuracy. To demonstrate the effectiveness of DropReef, we apply it to recent state-of-the-art sampling-based GNNs for training large-scale graphs, owing to the high precision of such models. With DropReef leveraged, the training efficiency of models can be greatly promoted. DropReef is highly compatible and is offline performed, benefiting the state-of-the-art sampling-based GNNs in the present and future to a significant extent.

We present a novel generative method for the creation of city-scale road layouts. While the output of recent methods is limited in both size of the covered area and diversity, our framework produces large traversable graphs of high quality consisting of vertices and edges representing complete street networks covering 400 square kilometers or more. While our framework can process general 2D embedded graphs, we focus on street networks due to the wide availability of training data. Our generative framework consists of a transformer decoder that is used in a sliding window manner to predict a field of indices, with each index encoding a representation of the local neighborhood. The semantics of each index is determined by a dictionary of context vectors. The index field is then input to a decoder to compute the street graph. Using data from OpenStreetMap, we train our system on whole cities and even across large countries such as the US, and finally compare it to the state of the art.

The Jobs and Skills Summit being convened this week in Canberra sets out to address the employment challenges that Australia currently faces. But even as we respond to today's skills shortage, we must also keep an eye on the long-term consequences of Artificial Intelligence (AI) and machine learning on jobs. Fantasies in which AIs advance to the point of doing all of our jobs are seductive in some tech circles. Elon Musk has long promised fully driverless cars that will be safer than cars driven by humans. When these vehicles do arrive, they may delete the human Uber driver.

For orthogonal multiple access (OMA) systems, the number of served user equipments (UEs) is limited to the number of available orthogonal resources. On the other hand, non-orthogonal multiple access (NOMA) schemes allow multiple UEs to use the same orthogonal resource. This extra degree of freedom introduces new challenges for resource allocation. Buffer state information (BSI), like the size and age of packets waiting for transmission, can be used to improve scheduling in OMA systems. In this paper, we investigate the impact of BSI on the performance of a centralized scheduler in an uplink multi-carrier NOMA scenario with UEs having various data rate and latency requirements. To handle the large combinatorial space of allocating UEs to the resources, we propose a novel scheduler based on actor-critic reinforcement learning incorporating BSI. Training and evaluation are carried out using Nokia's "wireless suite". We propose various novel techniques to both stabilize and speed up training. The proposed scheduler outperforms benchmark schedulers.

Bump hunting deals with finding in sample spaces meaningful data subsets known as bumps. These have traditionally been conceived as modal or concave regions in the graph of the underlying density function. We define an abstract bump construct based on curvature functionals of the probability density. Then, we explore several alternative characterizations involving derivatives up to second order. In particular, a suitable implementation of Good and Gaskins' original concave bumps is proposed in the multivariate case. Moreover, we bring to exploratory data analysis concepts like the mean curvature and the Laplacian that have produced good results in applied domains. Our methodology addresses the approximation of the curvature functional with a plug-in kernel density estimator. We provide theoretical results that assure the asymptotic consistency of bump boundaries in the Hausdorff distance with affordable convergence rates. We also present asymptotically valid and consistent confidence regions bounding curvature bumps. The theory is illustrated through several use cases in sports analytics with datasets from the NBA, MLB and NFL. We conclude that the different curvature instances effectively combine to generate insightful visualizations.