Multi-aspect controllable text generation is a more challenging and practical task than single-aspect control. Existing methods achieve complex multi-aspect control by fusing multiple controllers learned from single-aspect, but suffer from attribute degeneration caused by the mutual interference of these controllers. To address this, we provide observations on attribute fusion from a distributional perspective and propose to directly search for the intersection areas of multiple attribute distributions as their combination for generation. Our method first estimates the attribute space with an autoencoder structure. Afterward, we iteratively approach the intersections by jointly minimizing distances to points representing different attributes. Finally, we map them to attribute-relevant sentences with a prefix-tuning-based decoder. Experiments on the three-aspect control task, including sentiment, topic, and detoxification aspects, reveal that our method outperforms several strong baselines on attribute relevance and text quality and achieves the SOTA. Further analysis also supplies some explanatory support for the effectiveness of our approach.

Ensuring fairness of prediction-based decision making is based on statistical group fairness criteria. Which one of these criteria is the morally most appropriate one depends on the context, and its choice requires an ethical analysis. In this paper, we present a step-by-step procedure integrating three elements: (a) a framework for the moral assessment of what fairness means in a given context, based on the recently proposed general principle of "Fair equality of chances" (FEC) (b) a mapping of the assessment's results to established statistical group fairness criteria, and (c) a method for integrating the thus-defined fairness into optimal decision making. As a second contribution, we show new applications of the FEC principle and show that, with this extension, the FEC framework covers all types of group fairness criteria: independence, separation, and sufficiency. Third, we introduce an extended version of the FEC principle, which additionally allows accounting for morally irrelevant elements of the fairness assessment and links to well-known relaxations of the fairness criteria. This paper presents a framework to develop fair decision systems in a conceptually sound way, combining the moral and the computational elements of fair prediction-based decision-making in an integrated approach. Data and code to reproduce our results are available at https://github.com/joebaumann/fair-prediction-based-decision-making.

While vision-language pre-training model (VLP) has shown revolutionary improvements on various vision-language (V+L) tasks, the studies regarding its adversarial robustness remain largely unexplored. This paper studied the adversarial attack on popular VLP models and V+L tasks. First, we analyzed the performance of adversarial attacks under different settings. By examining the influence of different perturbed objects and attack targets, we concluded some key observations as guidance on both designing strong multimodal adversarial attack and constructing robust VLP models. Second, we proposed a novel multimodal attack method on the VLP models called Collaborative Multimodal Adversarial Attack (Co-Attack), which collectively carries out the attacks on the image modality and the text modality. Experimental results demonstrated that the proposed method achieves improved attack performances on different V+L downstream tasks and VLP models. The analysis observations and novel attack method hopefully provide new understanding into the adversarial robustness of VLP models, so as to contribute their safe and reliable deployment in more real-world scenarios. Code is available at https://github.com/adversarial-for-goodness/Co-Attack.

Despite the impressive performance of Artificial Intelligence (AI) systems, their robustness remains elusive and constitutes a key issue that impedes large-scale adoption. Robustness has been studied in many domains of AI, yet with different interpretations across domains and contexts. In this work, we systematically survey the recent progress to provide a reconciled terminology of concepts around AI robustness. We introduce three taxonomies to organize and describe the literature both from a fundamental and applied point of view: 1) robustness by methods and approaches in different phases of the machine learning pipeline; 2) robustness for specific model architectures, tasks, and systems; and in addition, 3) robustness assessment methodologies and insights, particularly the trade-offs with other trustworthiness properties. Finally, we identify and discuss research gaps and opportunities and give an outlook on the field. We highlight the central role of humans in evaluating and enhancing AI robustness, considering the necessary knowledge humans can provide, and discuss the need for better understanding practices and developing supportive tools in the future.

Fine-tuning over large pretrained language models (PLMs) has established many state-of-the-art results. Despite its superior performance, such fine-tuning can be unstable, resulting in significant variance in performance and potential risks for practical applications. Previous works have attributed such instability to the catastrophic forgetting problem in the top layers of PLMs, which indicates iteratively that fine-tuning layers in a top-down manner is a promising solution. In this paper, we first point out that this method does not always work out due to the different convergence speeds of different layers/modules. Inspired by this observation, we propose a simple component-wise gradient norm clipping method to adjust the convergence speed for different components. Experiment results demonstrate that our method achieves consistent improvements in terms of generalization performance, convergence speed, and training stability. The codebase can be found at https://github.com/yangalan123/FineTuningStability.

Demonstration-based learning has shown great potential in stimulating pretrained language models' ability under limited data scenario. Simply augmenting the input with some demonstrations can significantly improve performance on few-shot NER. However, why such demonstrations are beneficial for the learning process remains unclear since there is no explicit alignment between the demonstrations and the predictions. In this paper, we design pathological demonstrations by gradually removing intuitively useful information from the standard ones to take a deep dive of the robustness of demonstration-based sequence labeling and show that (1) demonstrations composed of random tokens still make the model a better few-shot learner; (2) the length of random demonstrations and the relevance of random tokens are the main factors affecting the performance; (3) demonstrations increase the confidence of model predictions on captured superficial patterns. We have publicly released our code at https://github.com/SALT-NLP/RobustDemo.

Since the 2004 DARPA Grand Challenge, the autonomous driving technology has witnessed nearly two decades of rapid development. Particularly, in recent years, with the application of new sensors and deep learning technologies extending to the autonomous field, the development of autonomous driving technology has continued to make breakthroughs. Thus, many carmakers and high-tech giants dedicated to research and system development of autonomous driving. However, as the foundation of autonomous driving, the deep learning technology faces many new security risks. The academic community has proposed deep learning countermeasures against the adversarial examples and AI backdoor, and has introduced them into the autonomous driving field for verification. Deep learning security matters to autonomous driving system security, and then matters to personal safety, which is an issue that deserves attention and research.This paper provides an summary of the concepts, developments and recent research in deep learning security technologies in autonomous driving. Firstly, we briefly introduce the deep learning framework and pipeline in the autonomous driving system, which mainly include the deep learning technologies and algorithms commonly used in this field. Moreover, we focus on the potential security threats of the deep learning based autonomous driving system in each functional layer in turn. We reviews the development of deep learning attack technologies to autonomous driving, investigates the State-of-the-Art algorithms, and reveals the potential risks. At last, we provides an outlook on deep learning security in the autonomous driving field and proposes recommendations for building a safe and trustworthy autonomous driving system.

When Quddus Pativada was 14, he wished that he had an app that could summarize his textbooks for him. Just five years later, Pativada has been there and done that -- earlier this year, he launched the AI-based app Kado, which turns photos, documents or PDFs into flash cards. Now, as the 19-year-old founder takes the stage for Startup Battlefield, he's looking to take his company, DigestAI, beyond flashcards to create an AI dialogue assistant that we can all carry around on our phones. "If we make learning truly easy and accessible, it's something you could do as soon as you open your phone," Pativada told TechCrunch. "We want to put a teacher in every single person's phone for every topic in the world."

Cleveland-born Samuel Alderson (1914-2005) created crash test dummies for the auto industry, drastically improving driver safety. Eleven additional people were killed in U.S. crashes involving vehicles that were using automated driving systems during a four-month period earlier this year, according to newly released government data, part of an alarming pattern of incidents linked to the technology. Ten of the deaths involved vehicles made by Tesla, though it is unclear from the National Highway Traffic Safety Administration's data whether the technology itself was at fault or whether driver error might have been responsible. The 11th death involved a Ford pickup truck. The deaths included four crashes involving motorcycles that occurred during the spring and summer: Two in Florida and one each in California and Utah.

Staring closely at the screen, Jordan Wright deftly picks out a barely distinguishable shape with his mouse, bringing to life a stark blue outline from a blur of overexposed features. It's a process similar to the automated tests that teach computers to distinguish humans from machines, by asking someone to identify traffic lights or stop signs in a picture known as a Captcha. Only in Wright's case, the shape turns out to be of a Tupolev Tu-160, a supersonic strategic heavy bomber, parked on a Russian base. The outline -- one of hundreds a day he picks out from satellite images -- is training an algorithm so a US intelligence agency can locate and identify Moscow's firepower in an automated flash. It's become a run-of-the-mill task for the 25-year-old, who describes himself as on the autism spectrum. Starting in the spring, Wright began working at Enabled Intelligence, a Virginia-based startup that works largely for US intelligence and other federal agencies.