Counterfactual explanations are an increasingly popular form of post hoc explanation due to their (i) applicability across problem domains, (ii) proposed legal compliance (e.g., with GDPR), and (iii) reliance on the contrastive nature of human explanation. Although counterfactual explanations are normally used to explain individual predictive-instances, we explore a novel use case in which groups of similar instances are explained in a collective fashion using ``group counterfactuals'' (e.g., to highlight a repeating pattern of illness in a group of patients). These group counterfactuals meet a human preference for coherent, broad explanations covering multiple events/instances. A novel, group-counterfactual algorithm is proposed to generate high-coverage explanations that are faithful to the to-be-explained model. This explanation strategy is also evaluated in a large, controlled user study (N=207), using objective (i.e., accuracy) and subjective (i.e., confidence, explanation satisfaction, and trust) psychological measures. The results show that group counterfactuals elicit modest but definite improvements in people's understanding of an AI system. The implications of these findings for counterfactual methods and for XAI are discussed.

AI-driven Action Quality Assessment (AQA) of sports videos can mimic Olympic judges to help score performances as a second opinion or for training. However, these AI methods are uninterpretable and do not justify their scores, which is important for algorithmic accountability. Indeed, to account for their decisions, instead of scoring subjectively, sports judges use a consistent set of criteria - rubric - on multiple actions in each performance sequence. Therefore, we propose IRIS to perform Interpretable Rubric-Informed Segmentation on action sequences for AQA. We investigated IRIS for scoring videos of figure skating performance. IRIS predicts (1) action segments, (2) technical element score differences of each segment relative to base scores, (3) multiple program component scores, and (4) the summed final score. In a modeling study, we found that IRIS performs better than non-interpretable, state-of-the-art models. In a formative user study, practicing figure skaters agreed with the rubric-informed explanations, found them useful, and trusted AI judgments more. This work highlights the importance of using judgment rubrics to account for AI decisions.

Remote Sensing Change Detection (RS-CD) aims to detect relevant changes from Multi-Temporal Remote Sensing Images (MT-RSIs), which aids in various RS applications such as land cover, land use, human development analysis, and disaster response. The performance of existing RS-CD methods is attributed to training on large annotated datasets. Furthermore, most of these models are less transferable in the sense that the trained model often performs very poorly when there is a domain gap between training and test datasets. This paper proposes an unsupervised CD method based on deep metric learning that can deal with both of these issues. Given an MT-RSI, the proposed method generates corresponding change probability map by iteratively optimizing an unsupervised CD loss without training it on a large dataset. Our unsupervised CD method consists of two interconnected deep networks, namely Deep-Change Probability Generator (D-CPG) and Deep-Feature Extractor (D-FE). The D-CPG is designed to predict change and no change probability maps for a given MT-RSI, while D-FE is used to extract deep features of MT-RSI that will be further used in the proposed unsupervised CD loss. We use transfer learning capability to initialize the parameters of D-FE. We iteratively optimize the parameters of D-CPG and D-FE for a given MT-RSI by minimizing the proposed unsupervised ``similarity-dissimilarity loss''. This loss is motivated by the principle of metric learning where we simultaneously maximize the distance between change pair-wise pixels while minimizing the distance between no-change pair-wise pixels in bi-temporal image domain and their deep feature domain. The experiments conducted on three CD datasets show that our unsupervised CD method achieves significant improvements over the state-of-the-art supervised and unsupervised CD methods. Code available at https://github.com/wgcban/Metric-CD

Recent years have seen growing adoption of AI-based decision-support systems (ADS) in homeless services, yet we know little about stakeholder desires and concerns surrounding their use. In this work, we aim to understand impacted stakeholders' perspectives on a deployed ADS that prioritizes scarce housing resources. We employed AI lifecycle comicboarding, an adapted version of the comicboarding method, to elicit stakeholder feedback and design ideas across various components of an AI system's design. We elicited feedback from county workers who operate the ADS daily, service providers whose work is directly impacted by the ADS, and unhoused individuals in the region. Our participants shared concerns and design suggestions around the AI system's overall objective, specific model design choices, dataset selection, and use in deployment. Our findings demonstrate that stakeholders, even without AI knowledge, can provide specific and critical feedback on an AI system's design and deployment, if empowered to do so.

Copyright protection for deep neural networks (DNNs) is an urgent need for AI corporations. To trace illegally distributed model copies, DNN watermarking is an emerging technique for embedding and verifying secret identity messages in the prediction behaviors or the model internals. Sacrificing less functionality and involving more knowledge about the target DNN, the latter branch called \textit{white-box DNN watermarking} is believed to be accurate, credible and secure against most known watermark removal attacks, with emerging research efforts in both the academy and the industry. In this paper, we present the first systematic study on how the mainstream white-box DNN watermarks are commonly vulnerable to neural structural obfuscation with \textit{dummy neurons}, a group of neurons which can be added to a target model but leave the model behavior invariant. Devising a comprehensive framework to automatically generate and inject dummy neurons with high stealthiness, our novel attack intensively modifies the architecture of the target model to inhibit the success of watermark verification. With extensive evaluation, our work for the first time shows that nine published watermarking schemes require amendments to their verification procedures.

Transformer-based language models (LMs) create hidden representations of their inputs at every layer, but only use final-layer representations for prediction. This obscures the internal decision-making process of the model and the utility of its intermediate representations. One way to elucidate this is to cast the hidden representations as final representations, bypassing the transformer computation in-between. In this work, we suggest a simple method for such casting, by using linear transformations. We show that our approach produces more accurate approximations than the prevailing practice of inspecting hidden representations from all layers in the space of the final layer. Moreover, in the context of language modeling, our method allows "peeking" into early layer representations of GPT-2 and BERT, showing that often LMs already predict the final output in early layers. We then demonstrate the practicality of our method to recent early exit strategies, showing that when aiming, for example, at retention of 95% accuracy, our approach saves additional 7.9% layers for GPT-2 and 5.4% layers for BERT, on top of the savings of the original approach. Last, we extend our method to linearly approximate sub-modules, finding that attention is most tolerant to this change.

Machine learning models can reach high performance on benchmark natural language processing (NLP) datasets but fail in more challenging settings. We study this issue when a pre-trained model learns dataset artifacts in natural language inference (NLI), the topic of studying the logical relationship between a pair of text sequences. We provide a variety of techniques for analyzing and locating dataset artifacts inside the crowdsourced Stanford Natural Language Inference (SNLI) corpus. We study the stylistic pattern of dataset artifacts in the SNLI. To mitigate dataset artifacts, we employ a unique multi-scale data augmentation technique with two distinct frameworks: a behavioral testing checklist at the sentence level and lexical synonym criteria at the word level. Specifically, our combination method enhances our model's resistance to perturbation testing, enabling it to continuously outperform the pre-trained baseline.

Recent work has proposed artificial intelligence (AI) models that can learn to decide whether to make a prediction for an instance of a task or to delegate it to a human by considering both parties' capabilities. In simulations with synthetically generated or context-independent human predictions, delegation can help improve the performance of human-AI teams -- compared to humans or the AI model completing the task alone. However, so far, it remains unclear how humans perform and how they perceive the task when they are aware that an AI model delegated task instances to them. In an experimental study with 196 participants, we show that task performance and task satisfaction improve through AI delegation, regardless of whether humans are aware of the delegation. Additionally, we identify humans' increased levels of self-efficacy as the underlying mechanism for these improvements in performance and satisfaction. Our findings provide initial evidence that allowing AI models to take over more management responsibilities can be an effective form of human-AI collaboration in workplaces.

The artificial intelligence research lab OpenAI on Tuesday launched the newest version of its language software, GPT-4, an advanced tool for analyzing images and mimicking human speech, pushing the technical and ethical boundaries of a rapidly proliferating wave of AI. OpenAI's earlier product, ChatGPT, captivated and unsettled the public with its uncanny ability to generate elegant writing, unleashing a viral wave of college essays, screenplays and conversations -- though it relied on an older generation of technology that hasn't been cutting-edge for more than a year. GPT-4, in contrast, is a state-of-the-art system capable of creating not just words but describing images in response to a person's simple written commands. When shown a photo of a boxing glove hanging over a wooden seesaw with a ball on one side, for instance, a person can ask what will happen if the glove drops, and GPT-4 will respond that it would hit the seesaw and cause the ball to fly up. The buzzy launch capped months of hype and anticipation over an AI program, known as a large language model, that early testers had claimed was remarkably advanced in its ability to reason and learn new things.

Surfers, swimmers and beachgoers face a hidden danger in the ocean: rip currents. These narrow channels of water can flow away from the shore at speeds up to 2.5 meters per second, making them one of the biggest safety risks for those enjoying the ocean. To help keep beachgoers safe, Christo Rautenbach, a coastal and estuarine physical processes scientist, has teamed up with the National Institute of Water and Atmospheric Research in New Zealand to develop a real-time rip current identification tool using deep learning. On this episode of the NVIDIA AI Podcast, host Noah Kravitz interviews Rautenbach about how AI can be used to identify rip currents and the potential for the tool to be used globally to help reduce the number of fatalities caused by rip currents. Developed in collaboration with Surf Lifesaving New Zealand, the rip current identification tool has achieved a detection rate of roughly 90% in trials.