Recent works propose using the discriminator of a GAN to filter out unrealistic samples of the generator. We generalize these ideas by introducing the implicit Metropolis-Hastings algorithm. For any implicit probabilistic model and a target distribution represented by a set of samples, implicit Metropolis-Hastings operates by learning a discriminator to estimate the density-ratio and then generating a chain of samples. Since the approximation of density ratio introduces an error on every step of the chain, it is crucial to analyze the stationary distribution of such chain. For that purpose, we present a theoretical result stating that the discriminator loss upper bounds the total variation distance between the target distribution and the stationary distribution. Finally, we validate the proposed algorithm both for independent and Markov proposals on CIFAR-10 and CelebA datasets.

This paper extensively evaluates the vulnerability of capsule networks to different adversarial attacks. Recent work suggests that these architectures are more robust towards adversarial attacks than other neural networks. However, our experiments show that capsule networks can be fooled as easily as convolutional neural networks.

The worst-case training principle that minimizes the maximal adversarial loss, also known as adversarial training (AT), has shown to be a state-of-the-art approach for enhancing adversarial robustness against norm-ball bounded input perturbations. Nonetheless, min-max optimization beyond the purpose of AT has not been rigorously explored in the research of adversarial attack and defense. In particular, given a set of risk sources (domains), minimizing the maximal loss induced from the domain set can be reformulated as a general min-max problem that is different from AT, since the maximization is taken over the probability simplex of the domain set. Examples of this general formulation include attacking model ensembles, devising universal perturbation to input samples or data transformations, and generalized AT over multiple norm-ball threat models. We show that these problems can be solved under a unified and theoretically principled min-max optimization framework. Our proposed approach leads to substantial performance improvement over the uniform averaging strategy in four different tasks. Moreover, we show how the self-adjusted weighting factors of the probability simplex from our proposed algorithms can be used to explain the importance of different attack and defense models.

Since 1950, when Alan Turing proposed what has since come to be called the Turing test, the ability of a machine to pass this test has established itself as the primary hallmark of general AI. To pass the test, a machine would have to be able to engage in dialogue in such a way that a human interrogator could not distinguish its behaviour from that of a human being. AI researchers have attempted to build machines that could meet this requirement, but they have so far failed. To pass the test, a machine would have to meet two conditions: (i) react appropriately to the variance in human dialogue and (ii) display a human-like personality and intentions. We argue, first, that it is for mathematical reasons impossible to program a machine which can master the enormously complex and constantly evolving pattern of variance which human dialogues contain. And second, that we do not know how to make machines that possess personality and intentions of the sort we find in humans. Since a Turing machine cannot master human dialogue behaviour, we conclude that a Turing machine also cannot possess what is called ``general'' Artificial Intelligence. We do, however, acknowledge the potential of Turing machines to master dialogue behaviour in highly restricted contexts, where what is called ``narrow'' AI can still be of considerable utility.

We propose a novel method for exploiting the semantic structure of text to answer multiple-choice questions. The approach is especially suitable for domains that require reasoning over a diverse set of linguistic constructs but have limited training data. To address these challenges, we present the first system, to the best of our knowledge, that reasons over a wide range of semantic abstractions of the text, which are derived using off-the-shelf, general-purpose, pre-trained natural language modules such as semantic role labelers, coreference resolvers, and dependency parsers. Representing multiple abstractions as a family of graphs, we translate question answering (QA) into a search for an optimal subgraph that satisfies certain global and local properties. This formulation generalizes several prior structured QA systems. Our system, SEMANTICILP, demonstrates strong performance on two domains simultaneously. In particular, on a collection of challenging science QA datasets, it outperforms various state-of-the-art approaches, including neural models, broad coverage information retrieval, and specialized techniques using structured knowledge bases, by 2%-6%.

Semantically controlled neural response generation on limited-domain has achieved great performance. However, moving towards multi-domain large-scale scenarios are shown to be difficult because the possible combinations of semantic inputs grow exponentially with the number of domains. To alleviate such scalability issue, we exploit the structure of dialog acts to build a multi-layer hierarchical graph, where each act is represented as a root-to-leaf route on the graph. Then, we incorporate such graph structure prior as an inductive bias to build a hierarchical disentangled self-attention network, where we disentangle attention heads to model designated nodes on the dialog act graph. By activating different (disentangled) heads at each layer, combinatorially many dialog act semantics can be modeled to control the neural response generation. On the large-scale Multi-Domain-WOZ dataset, our model can yield a significant improvement over the baselines on various automatic and human evaluation metrics.

Image-processing algorithms to smartly identify, caption and moderate your pictures. Convert spoken audio into text, use voice for verification, or add speaker recognition to your app. Allow your apps to process natural language with pre-built scripts, evaluate sentiment and learn how to recognize what users want. Map complex information and data in order to solve tasks such as intelligent recommendations and semantic search. Add Google or Bing Search APIs to your apps and harness the ability to comb billions of webpages, images, videos, news and much more.

The growing use of artificial intelligence in sensitive areas, including for hiring, criminal justice, and healthcare, has stirred a debate about bias and fairness. Yet human decision making in these and other domains can also be flawed, shaped by individual and societal biases that are often unconscious. Will AI's decisions be less biased than human ones? Or will AI make these problems worse? Will AI's decisions be less biased than human ones?

Why are coal plants in the U.S. and Europe closing at an accelerating rate, while in Asia, coal consumption went up and helped fuel an overall 1.7% year-on-year increase in global carbon emissions? Part of the reason coal continues to grow in countries like China and India is that in these areas, unlike in the U.S., emissions data can be shoddy or hard to acquire. Without accurate information it is harder to hold facilities accountable and keep them in line with meeting emission reduction targets. To address this situation, we are partnering with WattTime and the World Resources Institute (WRI), to launch a new project which will use satellite imagery to quantify carbon emissions from every major power plant across the world. This effort is being funded as one of 20 projects in the Google AI Impact Challenge.

With more access to data and growing computing power, artificial intelligence (AI) is becoming increasingly powerful. But for it to be effective and meaningful, we must embrace people-first artificial intelligence strategies, according to Soumitra Dutta, professor of operations, technology, and information management at the Cornell SC Johnson College of Business. "There has to be a human agency-first kind of principle that lets people feel empowered about how to make decisions and how to use AI systems to support their decision-making," notes Dutta. Knowledge@Wharton interviewed him at a recent conference on artificial intelligence and machine learning in the financial industry, organized in New York City by the SWIFT Institute in collaboration with Cornell's SC Johnson College of Business. In this conversation, Dutta discusses some myths around AI, what it means to have a people-first artificial intelligence strategy, why it is important, and how we can overcome the challenges in realizing this vision. An edited transcript of the conversation follows. Knowledge@Wharton: What are some of the biggest myths about AI, especially as they relate to financial services?