Blackbox transfer attacks for image classifiers have been extensively studied in recent years. In contrast, little progress has been made on transfer attacks for object detectors. Object detectors take a holistic view of the image and the detection of one object (or lack thereof) often depends on other objects in the scene. This makes such detectors inherently context-aware and adversarial attacks in this space are more challenging than those targeting image classifiers. In this paper, we present a new approach to generate context-aware attacks for object detectors. We show that by using co-occurrence of objects and their relative locations and sizes as context information, we can successfully generate targeted mis-categorization attacks that achieve higher transfer success rates on blackbox object detectors than the state-of-the-art. We test our approach on a variety of object detectors with images from PASCAL VOC and MS COCO datasets and demonstrate up to $20$ percentage points improvement in performance compared to the other state-of-the-art methods.

Due to their high energy intensity, buildings play a major role in the current worldwide energy transition. Building models are ubiquitous since they are needed at each stage of the life of buildings, i.e. for design, retrofitting, and control operations. Classical white-box models, based on physical equations, are bound to follow the laws of physics but the specific design of their underlying structure might hinder their expressiveness and hence their accuracy. On the other hand, black-box models are better suited to capture nonlinear building dynamics and thus can often achieve better accuracy, but they require a lot of data and might not follow the laws of physics, a problem that is particularly common for neural network (NN) models. To counter this known generalization issue, physics-informed NNs have recently been introduced, where researchers introduce prior knowledge in the structure of NNs to ground them in known underlying physical laws and avoid classical NN generalization issues. In this work, we present a novel physics-informed NN architecture, dubbed Physically Consistent NN (PCNN), which only requires past operational data and no engineering overhead, including prior knowledge in a linear module running in parallel to a classical NN. We formally prove that such networks are physically consistent -- by design and even on unseen data -- with respect to different control inputs and temperatures outside and in neighboring zones. We demonstrate their performance on a case study, where the PCNN attains an accuracy up to $50\%$ better than a classical physics-based resistance-capacitance model on $3$-day long prediction horizons. Furthermore, despite their constrained structure, PCNNs attain similar performance to classical NNs on the validation data, overfitting the training data less and retaining high expressiveness to tackle the generalization issue.

Single-agent reinforcement learning algorithms in a multi-agent environment are inadequate for fostering cooperation. If intelligent agents are to interact and work together to solve complex problems, methods that counter non-cooperative behavior are needed to facilitate the training of multiple agents. This is the goal of cooperative AI. Recent work in adversarial machine learning, however, shows that models (e.g., image classifiers) can be easily deceived into making incorrect decisions. In addition, some past research in cooperative AI has relied on new notions of representations, like public beliefs, to accelerate the learning of optimally cooperative behavior. Hence, cooperative AI might introduce new weaknesses not investigated in previous machine learning research. In this paper, our contributions include: (1) arguing that three algorithms inspired by human-like social intelligence introduce new vulnerabilities, unique to cooperative AI, that adversaries can exploit, and (2) an experiment showing that simple, adversarial perturbations on the agents' beliefs can negatively impact performance. This evidence points to the possibility that formal representations of social behavior are vulnerable to adversarial attacks.

Job automation, algorithmic bias, and technological development are the first thoughts that spring to mind when we think of Artificial Intelligence. But at the same time, AI can be used in many fun and interesting ways. Here, we discuss ten fun AI tools that you must try out. Besides being a great way to kill boredom, they demonstrate how advanced AI has already become. Semantris is one of the many Google-powered AI experiments.

Sen. Joni Ernst, R-Iowa, and former under Secretary of Defense for policy Michèle Flournoy discuss possible actions to take if Russia invades Ukraine. This is a rush transcript of "Fox News Sunday" on December 5, 2021. This copy may not be in its final form and may be updated. President Biden and Russia's Vladimir Putin will hold a superpower phone JOE BIDEN, PRESIDENT OF THE UNITED STATES: I don't accept anybody's red We'll discuss the standoff with Senate Armed Services Committee member Joni Just how much of a threat is China? We'll talk about how to keep law and order in space with the vice chief of So, we need to be ready. U.S. faces around the world.

Rep. Elise Stefanik, R-NY, reacts to the former CNN anchor being fired over his role in former Gov. Andrew Cuomo's sexual harassment scandal. White House chief of staff Ronald Klain confused readers Sunday as he promoted a Washington Post op-ed that argued President Biden gets worse media treatment than his predecessor, former President Trump, whose verbal duels with the press were weekly staples during his four-year residency at 1600 Penn. "For your consideration," Klain tweeted with a link to the op-ed from Dana Millbank, titled, "The media treats Biden as badly as - or worse than - Trump. WHITE HOUSE'S RON KLAIN PANNED FOR RETWEETING POST ON'ULTIMATE WORK AROUND' FOR FEDERAL VACCINE MANDATE Millbank's "proof" was research from Forge.ai, a data analytics unit of the information company FiscalNote. The study used algorithms focused on adjectives and their placement in articles - more than 200,000 of them - to rate the coverage Biden received in the first 11 months of 2021 and the coverage Trump got in the first 11 months of 2020. The process was referred to as "sentiment analysis." "My colleagues in the media are serving as accessories to the murder of democracy," Millbank said. "Too many journalists are caught in a mindless neutrality between democracy and its saboteurs, between fact and fiction.

Civil rights groups argue that facial recognition technology is error-prone, misidentifying women and minorities at higher rates than it does white men and sometimes leading to false arrests. Clearview AI has gotten the green light on a federal patent for its facial recognition technology -- an award that the company says is the first to cover a so-called "search engine for faces" that crawls the internet to find matches. Clearview's software -- which scrapes public images from social media to help law enforcement match images in government databases or surveillance footage -- has long faced fire from privacy advocates who say it uses people's faces without their knowledge or consent. Civil rights groups also argue that facial recognition technology is generally error-prone, misidentifying women and minorities at higher rates than it does white men and sometimes leading to false arrests.

One issue that is not receiving enough attention is that of cyberattacks evolving from being performed by humans against other humans through the use of technology, to computers attacking computers, with little human involvement. Another important AI and cybersecurity concern is that AI systems can be hacked by feeding them bad data from which to learn – the systems themselves do not need to be compromised to render them impotent or worse. The first is to prevent the AI from making incorrect observations and decisions because it suffers from a lack of complete data, or biased data, for the matters which it is supposed to address. Ensuring that systems are created and tested by diverse teams, as well as that during learning phases systems are fed sufficiently diverse data, can help address such concerns. The second issue is that AIs may make decisions that society views as inappropriately biased, but which the AI considers being both correct and integral to achieving maximum performance.

This work explores the potency of stochastic competition-based activations, namely Stochastic Local Winner-Takes-All (LWTA), against powerful (gradient-based) white-box and black-box adversarial attacks; we especially focus on Adversarial Training settings. In our work, we replace the conventional ReLU-based nonlinearities with blocks comprising locally and stochastically competing linear units. The output of each network layer now yields a sparse output, depending on the outcome of winner sampling in each block. We rely on the Variational Bayesian framework for training and inference; we incorporate conventional PGD-based adversarial training arguments to increase the overall adversarial robustness. As we experimentally show, the arising networks yield state-of-the-art robustness against powerful adversarial attacks while retaining very high classification rate in the benign case.

Data-driven forecasts of air quality have recently achieved more accurate short-term predictions. Despite their success, most of the current data-driven solutions lack proper quantifications of model uncertainty that communicate how much to trust the forecasts. Recently, several practical tools to estimate uncertainty have been developed in probabilistic deep learning. However, there have not been empirical applications and extensive comparisons of these tools in the domain of air quality forecasts. Therefore, this work applies state-of-the-art techniques of uncertainty quantification in a real-world setting of air quality forecasts. Through extensive experiments, we describe training probabilistic models and evaluate their predictive uncertainties based on empirical performance, reliability of confidence estimate, and practical applicability. We also propose improving these models using "free" adversarial training and exploiting temporal and spatial correlation inherent in air quality data. Our experiments demonstrate that the proposed models perform better than previous works in quantifying uncertainty in data-driven air quality forecasts. Overall, Bayesian neural networks provide a more reliable uncertainty estimate but can be challenging to implement and scale. Other scalable methods, such as deep ensemble, Monte Carlo (MC) dropout, and stochastic weight averaging-Gaussian (SWAG), can perform well if applied correctly but with different tradeoffs and slight variations in performance metrics. Finally, our results show the practical impact of uncertainty estimation and demonstrate that, indeed, probabilistic models are more suitable for making informed decisions. Code and dataset are available at \url{https://github.com/Abdulmajid-Murad/deep_probabilistic_forecast}