When looking for help with a task at work, people turn to those best able to do the job. Right? Wrong. New research shows that work partners tend to be chosen not for ability but for likability. Drawing from their study encompassing 10,000 work relationships in five organizations, the authors have classified work partners into four archetypes: the competent jerk, who knows a lot but is unpleasant; the lovable fool, who doesn't know much but is a delight; the lovable star, who's both smart and likable; and the incompetent jerk, who.. .well, that's self-explanatory. Of course, everybody wants to work with the lovable star, and nobody wants to work with the incompetent jerk. More interesting is that people prefer the lovable fool over the competent jerk. That has big implications for every organization, as both of these types often represent missed opportunities. Because they are liked by a disproportionate number of people, lovable fools can bridge gaps between diverse groups that might not otherwise interact. But their networking skills are often developed at the expense of job performance, which can make these employees underappreciated and vulnerable to downsizing. To get the most out of them, managers need to protect them and put them in positions that don't waste their bridge-building talents. As for the competent jerks, too often their expertise goes untapped by people who just can't put up with them. But many can be socialized through coaching or by being made accountable for bad behavior. Others may need to display their competence in more isolated settings. Intriguingly, managers aren't limited to leveraging people that others like and changing those that others loathe. They also can create situations in which people are more apt to like one another, whatever their individual qualities.
An adversarial example is an example that has been adjusted to produce a wrong label when presented to a system at test time. To date, adversarial example constructions have been demonstrated for classifiers, but not for detectors. If adversarial examples that could fool a detector exist, they could be used to (for example) maliciously create security hazards on roads populated with smart vehicles. In this paper, we demonstrate a construction that successfully fools two standard detectors, Faster RCNN and YOLO. The existence of such examples is surprising, as attacking a classifier is very different from attacking a detector, and that the structure of detectors - which must search for their own bounding box, and which cannot estimate that box very accurately - makes it quite likely that adversarial patterns are strongly disrupted. We show that our construction produces adversarial examples that generalize well across sequences digitally, even though large perturbations are needed. We also show that our construction yields physical objects that are adversarial.
Recent work has shown that neural network-based vision classifiers exhibit a significant vulnerability to misclassifications caused by imperceptible but adversarial perturbations of their inputs. These perturbations, however, are purely pixel-wise and built out of loss function gradients of either the attacked model or its surrogate. As a result, they tend to be contrived and look pretty artificial. This might suggest that such vulnerability to slight input perturbations can only arise in a truly adversarial setting and thus is unlikely to be an issue in more natural contexts. In this paper, we provide evidence that such belief might be incorrect. We demonstrate that significantly simpler, and more likely to occur naturally, transformations of the input - namely, rotations and translations alone, suffice to significantly degrade the classification performance of neural network-based vision models across a spectrum of datasets. This remains to be the case even when these models are trained using appropriate data augmentation. Finding such fooling transformations does not require having any special access to the model - just trying out a small number of random rotation and translation combinations already has a significant effect. These findings suggest that our current neural network-based vision models might not be as reliable as we tend to assume. Finally, we consider a new class of perturbations that combines rotations and translations with the standard pixel-wise attacks. We observe that these two types of input transformations are, in a sense, orthogonal to each other. Their effect on the performance of the model seems to be additive, while robustness to one type does not seem to affect the robustness to the other type. This suggests that this combined class of transformations is a more complete notion of similarity in the context of adversarial robustness of vision models.
As machine learning black boxes are increasingly being deployed in domains such as healthcare and criminal justice, there is growing emphasis on building tools and techniques for explaining these black boxes in an interpretable manner. Such explanations are being leveraged by domain experts to diagnose systematic errors and underlying biases of black boxes. In this paper, we demonstrate that post hoc explanations techniques that rely on input perturbations, such as LIME and SHAP, are not reliable. Specifically, we propose a novel scaffolding technique that effectively hides the biases of any given classifier by allowing an adversarial entity to craft an arbitrary desired explanation. Our approach can be used to scaffold any biased classifier in such a way that its predictions on the input data distribution still remain biased, but the post hoc explanations of the scaffolded classifier look innocuous. Using extensive evaluation with multiple real-world datasets (including COMPAS), we demonstrate how extremely biased (racist) classifiers crafted by our framework can easily fool popular explanation techniques such as LIME and SHAP into generating innocuous explanations which do not reflect the underlying biases.
Generating adversarial examples is a critical step for evaluating and improving the robustness of learning machines. So far, most existing methods only work for classification and are not designed to alter the true performance measure of the problem at hand. We introduce a novel flexible approach named Houdini for generating adversarial examples specifically tailored for the final performance measure of the task considered, be it combinatorial and non-decomposable. We successfully apply Houdini to a range of applications such as speech recognition, pose estimation and semantic segmentation. In all cases, the attacks based on Houdini achieve higher success rate than those based on the traditional surrogates used to train the models while using a less perceptible adversarial perturbation.
"In this study, Sergey A. Ivanov examines the phenomenon of holy foolery from a cultural standpoint. Adopting a diachronic approach, he identifies the prerequisites for holy foolery, considers the way it was shaped in the religious mind, and follows the emergence of the first hagiographic texts describing these paradoxical saints. In exploring the subsequent tradition of holy foolery, he shows how the genuinely insane could be promoted to the ranks of holy fools, as new generations of hagiographers attempted to read insanity in the light of the established canon. He also uncovers a large number of Byzantine and Old Russian sources dealing with holy fools who, for various reasons, were not entered into the official lists of saints."--BOOK JACKET.
State-of-the-art deep neural networks have achieved impressive results on many image classification tasks. However, these same architectures have been shown to be unstable to small, well sought, perturbations of the images. Despite the importance of this phenomenon, no effective methods have been proposed to accurately compute the robustness of state-of-the-art deep classifiers to such perturbations on large-scale datasets. In this paper, we fill this gap and propose the DeepFool algorithm to efficiently compute perturbations that fool deep networks, and thus reliably quantify the robustness of these classifiers. Extensive experimental results show that our approach outperforms recent methods in the task of computing adversarial perturbations and making classifiers more robust.
Studies demonstrate that elite athletes are able to extract kinematic information of observed domain-specific actions to predict their future course. Little is known, however, on the perceptuo-motor processes and neural correlates of the athletes' ability to predict fooling actions. Combining psychophysics and transcranial magnetic stimulation, we explored the impact of motor and perceptual expertise on the ability to predict the fate of observed actual or fake soccer penalty kicks. We manipulated the congruence between the model's body kinematics and the subsequent ball trajectory and investigated the prediction performance and cortico-spinal reactivity of expert kickers, goalkeepers, and novices. Kickers and goalkeepers outperformed novices by anticipating the actual kick direction from the model's initial body movements. However, kickers were more often fooled than goalkeepers and novices in cases of incongruent actions. Congruent and incongruent actions engendered a comparable facilitation of kickers' lower limb motor representation, but their neurophysiological response was correlated with their greater susceptibility to be fooled. Moreover, when compared with actual actions, motor facilitation for incongruent actions was lower among goalkeepers and higher among novices. Thus, responding to fooling actions requires updation of simulative motor representations of others' actions and is facilitated by visual rather than by motor expertise.
Deep neural networks (DNNs) have recently been achieving state-of-the-art performance on a variety of pattern-recognition tasks, most notably visual classification problems. Given that DNNs are now able to classify objects in images with near-human-level performance, questions naturally arise as to what differences remain between computer and human vision. A recent study [30] revealed that changing an image (e.g. of a lion) in a way imperceptible to humans can cause a DNN to label the image as something else entirely (e.g. mislabeling a lion a library). Here we show a related result: it is easy to produce images that are completely unrecognizable to humans, but that state-of-the-art DNNs believe to be recognizable objects with 99.99% confidence (e.g. labeling with certainty that white noise static is a lion). Specifically, we take convolutional neural networks trained to perform well on either the ImageNet or MNIST datasets and then find images with evolutionary algorithms or gradient ascent that DNNs label with high confidence as belonging to each dataset class. It is possible to produce images totally unrecognizable to human eyes that DNNs believe with near certainty are familiar objects, which we call “fooling images” (more generally, fooling examples). Our results shed light on interesting differences between human vision and current DNNs, and raise questions about the generality of DNN computer vision.
This paper explores the role of the sage-fool as means of creating a counter-vailing power against the regressive forces inherent in leadership, in other words, to reinforce the leader's capacity for reality testing. After a brief review of leadership theories, some of the salient characteristics of the role of the sage-fool are reviewed, taking a historical perspective. It is subsequently demonstrated how the fool, in playing the role of mediator between leader and followers, brings to the surface certain conflictual themes and thereby allows both parties to deal with the issues at hand. In this context, the role of humor and joking relationships are examined. Finally, two case examples of the fool in an organizational setting are presented.
Nowadays, cameras equipped with AI systems can capture and analyze images to detect people automatically. However, the AI system can make mistakes when receiving deliberately designed patterns in the real world, i.e., physical adversarial examples. Prior works have shown that it is possible to print adversarial patches on clothes to evade DNN-based person detectors. However, these adversarial examples could have catastrophic drops in the attack success rate when the viewing angle (i.e., the camera's angle towards the object) changes. To perform a multi-angle attack, we propose Adversarial Texture (AdvTexture). AdvTexture can cover clothes with arbitrary shapes so that people wearing such clothes can hide from person detectors from different viewing angles. We propose a generative method, named Toroidal-Cropping-based Expandable Generative Attack (TC-EGA), to craft AdvTexture with repetitive structures. We printed several pieces of cloth with AdvTexure and then made T-shirts, skirts, and dresses in the physical world. Experiments showed that these clothes could fool person detectors in the physical world.
Adversarial attacks on machine learning models have seen increasing interest in the past years. By making only subtle changes to the input of a convolutional neural network, the output of the network can be swayed to output a completely different result. The first attacks did this by changing pixel values of an input image slightly to fool a classifier to output the wrong class. Other approaches have tried to learn ``patches'' that can be applied to an object to fool detectors and classifiers. Some of these approaches have also shown that these attacks are feasible in the real-world, i.e. by modifying an object and filming it with a video camera. However, all of these approaches target classes that contain almost no intra-class variety (e.g. stop signs). The known structure of the object is then used to generate an adversarial patch on top of it. In this paper, we present an approach to generate adversarial patches to targets with lots of intra-class variety, namely persons. The goal is to generate a patch that is able successfully hide a person from a person detector. An attack that could for instance be used maliciously to circumvent surveillance systems, intruders can sneak around undetected by holding a small cardboard plate in front of their body aimed towards the surveilance camera. From our results we can see that our system is able significantly lower the accuracy of a person detector. Our approach also functions well in real-life scenarios where the patch is filmed by a camera. To the best of our knowledge we are the first to attempt this kind of attack on targets with a high level of intra-class variety like persons.
We show that any distribution on $\{-1,+1\}^n$ that is k-wise independent fools any halfspace (or linear threshold function) $h:\{-1,+1\}^n\to\{-1,+1\}$, i.e., any function of the form $h(x)=\operatorname{sign}(\sum_{i=1}^{n}w_{i}x_{i}-\theta)$, where the $w_1,\dots,w_n$ and $\theta$ are arbitrary real numbers, with error $\epsilon$ for $k=O(\epsilon^{-2}\log^2(1/\epsilon))$. Our result is tight up to $\log(1/\epsilon)$ factors. Using standard constructions of k-wise independent distributions, we obtain the first explicit pseudorandom generators $G:\{-1,+1\}^s\to\{-1,+1\}^n$ that fool halfspaces. Specifically, we fool halfspaces with error $\epsilon$ and seed length $s=k\cdot\log n=O(\log n\cdot\epsilon^{-2}\log^2(1/\epsilon))$. Our approach combines classical tools from real approximation theory with structural results on halfspaces by Servedio [Comput. Complexity, 16 (2007), pp. 180–209].
The fool is a social type having certain definable roles and a special status and function in group life. Fools represent departures from group norms of propriety which are subject to the sanction of ridicule. Fool-making is a continuous social process and operates to enforce propierty and to adjust status.
In this fascinating interpretation of contemporary culture and theology, Harvey Cox examines both the loss and reemergence of festivity and fantasy in Western civilization. He evaluates both processes from a theological perspective, defining festivity as the capacity for genuine revelry and joyous celebration and defining fantasy as the faculty for envisioning radically alternative life situations. He asserts that both are absolutely vital to contemporary human life and faith; both are a precondition for genuine social transformation. In a success and money-oriented society we need a rebirth of unapologetically unproductive festivity and expressive celebration. In an age that has quarantined parody and separated politics from imagination, we need a renaissance of social fantasy. It has been said over and over again that affluent Western man has been gaining the whole world while losing his soul. In the face of this Mr. Cox affirms the possibility and necessity of a resurgence of hope, celebration, liberation, and experimentation. The medieval Feast of Fools, from which he has taken his title, symbolizes both the problem and the process. Centuries ago it provided an opportunity for the choirboy to play bishop and for serious townsfolk to mock the stately rituals of church and court. The eventual disappearance of the custom in the sixteenth century, unlamented if not welcomed by those in authority, illustrates the concerns of this provocative and controversial essay. Mr. Cox does not propose that a medieval practice should be revived, but he does argue for a rebirth in our own cultural idiom of what was right and good about the Feast of Fools. It is likely that this book will become significant in wide circles. It speaks directly to such contemporary movements as the theology of hope, the rapidly disappearing radical theology, and the theology of culture. For many it will provide a new perspective on the renewal of religious life and the secular search for religious experience. For others it will function as a window into the experimental laboratories of the so-called "underground church." For everyone it is a refreshing encounter with a wholly new set of perceptive observations about the problems that plague us.
As machine learning black boxes are increasingly being deployed in domains such as healthcare and criminal justice, there is growing emphasis on building tools and techniques for explaining these black boxes in an interpretable manner. Such explanations are being leveraged by domain experts to diagnose systematic errors and underlying biases of black boxes. In this paper, we demonstrate that post hoc explanations techniques that rely on input perturbations, such as LIME and SHAP, are not reliable. Specifically, we propose a novel scaffolding technique that effectively hides the biases of any given classifier by allowing an adversarial entity to craft an arbitrary desired explanation. Our approach can be used to scaffold any biased classifier in such a way that its predictions on the input data distribution still remain biased, but the post hoc explanations of the scaffolded classifier look innocuous. Using extensive evaluation with multiple real world datasets (including COMPAS), we demonstrate how extremely biased (racist) classifiers crafted by our framework can easily fool popular explanation techniques such as LIME and SHAP into generating innocuous explanations which do not reflect the underlying biases.
En 1881, Edgeworth posait comme premier principe de l'economie le fait que chaque agent a comme motivation unique son interet propre. Cette conception egoiste de l'homme a penetre la theorie economique et y persiste. L'A. demontre l'inconsistance des concepts qui en sont les points nodaux: rationalite de la conduite qui consiste a choisir selon l'ordre de ses preferences ou selon le critere de l'engagement financier ou de l'utilite.
This volume presents three major social types in American society-heroes, villains, and fools-as models for American behaviour. Approaching these models primarily through language, Orrin E. Klapp explores what they may suggest about Americans as a people. Rather than study people, the author describes abstract types named and embedded in popular language. These social types are important symbols; and a way to attack a symbol is by identifying its meaning in various contexts. He further argues that the language surrounding heroes, villains, and fools reveals a social structure. We may not escape being ascribed a type, but we do have a choice of type. Known more commonly as finding oneself, we can manipulate cues-with dress, facial expressions, style of life, or conspicuous public roles-to build an identity. This classic study has serious contemporary implications. For a public figure, an inevitable result of the typing process is the development of at least two selves, the public and the private. When the book originally appeared in 1962, the struggle to balance two images generally only plagued celebrities and politicians. Today, social media offers everyone the opportunity to develop an online persona. This volume will be of interest to sociologists as well as anyone who has a Facebook account.
暂无摘要（点击查看原文获取完整内容）