This paper presents an algorithmic framework for learning robust policies in asymmetric imperfect-information games, where the joint reward could depend on the uncertain opponent type (a private information known only to the opponent itself and its ally). In order to maximize the reward, the protagonist agent has to infer the opponent type through agent modeling. We use multiagent reinforcement learning (MARL) to learn opponent models through self-play, which captures the full strategy interaction and reasoning between agents. However, agent policies learned from self-play can suffer from mutual overfitting. Ensemble training methods can be used to improve the robustness of agent policy against different opponents, but it also significantly increases the computational overhead. In order to achieve a good trade-off between the robustness of the learned policy and the computation complexity, we propose to train a separate opponent policy against the protagonist agent for evaluation purposes. The reward achieved by this opponent is a noisy measure of the robustness of the protagonist agent policy due to the intrinsic stochastic nature of a reinforcement learner. To handle this stochasticity, we apply a stochastic optimization scheme to dynamically update the opponent ensemble to optimize an objective function that strikes a balance between robustness and computation complexity. We empirically show that, under the same limited computational budget, the proposed method results in more robust policy learning than standard ensemble training.

Native speakers can judge whether a sentence is an acceptable instance of their language. Acceptability provides a means of evaluating whether computational language models are processing language in a human-like manner. We test the ability of computational language models, simple language features, and word embeddings to predict native English speakers judgments of acceptability on English-language essays written by non-native speakers. We find that much of the sentence acceptability variance can be captured by a combination of features including misspellings, word order, and word similarity (Pearson's r = 0.494). While predictive neural models fit acceptability judgments well (r = 0.527), we find that a 4-gram model with statistical smoothing is just as good (r = 0.528). Thanks to incorporating a count of misspellings, our 4-gram model surpasses both the previous unsupervised state-of-the art (Lau et al., 2015; r = 0.472), and the average non-expert native speaker (r = 0.46). Our results demonstrate that acceptability is well captured by n-gram statistics and simple language features.

From a set of technical drawings and expert knowledge, we automatically learn a parser to interpret such a drawing. This enables automatic reasoning and learning on top of a large database of technical drawings. In this work, we develop a similarity based search algorithm to help engineers and designers find or complete designs more easily and flexibly. This is part of an ongoing effort to build an automated engineering assistant. The proposed methods make use of both neural methods to learn to interpret images, and symbolic methods to learn to interpret the structure in the technical drawing and incorporate expert knowledge.

Constraint satisfaction problems form an important and wide class of combinatorial search and optimization problems with many applications in AI and other areas. We introduce a recurrent neural network architecture RUN-CSP (Recurrent Unsupervised Neural Network for Constraint Satisfaction Problems) to train message passing networks solving binary constraint satisfaction problems (CSPs) or their optimization versions (Max-CSP). The architecture is universal in the sense that it works for all binary CSPs: depending on the constraint language, we can automtically design a loss function, which is then used to train generic neural nets. In this paper, we experimentally evaluate our approach for the 3-colorability problem (3-Col) and its optimization version (Max-3-Col) and for the maximum 2-satisfiability problem (Max-2-Sat). We also extend the framework to work for related optimization problems such as the maximum independent set problem (Max-IS). Training is unsupervised, we train the network on arbitrary (unlabeled) instances of the problems. Moreover, we experimentally show that it suffices to train on relatively small instances; the resulting message passing network will perform well on much larger instances (at least 10-times larger).

Our goal is to better comprehend procedural text, e.g., a paragraph about photosynthesis, by not only predicting what happens, but why some actions need to happen before others. Our approach builds on a prior process comprehension framework for predicting actions' effects, to also identify subsequent steps that those effects enable. We present our new model (XPAD) that biases effect predictions towards those that (1) explain more of the actions in the paragraph and (2) are more plausible with respect to background knowledge. We also extend an existing benchmark dataset for procedural text comprehension, ProPara, by adding the new task of explaining actions by predicting their dependencies. We find that XPAD significantly outperforms prior systems on this task, while maintaining the performance on the original task in ProPara. The dataset is available at http://data.allenai.org/propara

The container pre-marshalling problem (CPMP) is concerned with the re-ordering of containers in container terminals during off-peak times so that containers can be quickly retrieved when the port is busy. The problem has received significant attention in the literature and is addressed by a large number of exact and heuristic methods. Existing methods for the CPMP heavily rely on problem-specific components (e.g., proven lower bounds) that need to be developed by domain experts with knowledge of optimization techniques and a deep understanding of the problem at hand. With the goal to automate the costly and time-intensive design of heuristics for the CPMP, we propose a new method called Deep Learning Heuristic Tree Search (DLTS). It uses deep neural networks to learn solution strategies and lower bounds customized to the CPMP solely through analyzing existing (near-) optimal solutions to CPMP instances. The networks are then integrated into a tree search procedure to decide which branch to choose next and to prune the search tree. DLTS produces the highest quality heuristic solutions to the CPMP to date with gaps to optimality below 2% on real-world sized instances.

Nvidia Corp. is upping its artificial intelligence game with the release of a new version of its TensorRT software platform for high-performance deep learning inference. TensorRT is a platform that combines a high-performance deep learning inference optimizer with a runtime that delivers low-latency, high-throughput inference for AI applications. Inference is an important aspect of AI. Whereas AI training relates to the development of an algorithm's ability to understand a data set, inference refers to its ability to act on that data to infer answers to specific queries. The latest version brings with it some dramatic improvements on the performance side.

Earth is enormous, and while humans have done a decent job of being able to map out the boundaries of countries and states, the roads in our cities and the location of geological sightseeing destinations, there remains a lot of the world that isn't precisely figured out. But a new project from Wenwen Li, associate professor in the School of Geographical Sciences and Urban Planning, aims to learn more about our world and its varying terrain by applying artificial intelligence. Artificial intelligence, or AI, has already made an indelible impact in daily life. From knowing our commutes or being able to suggest new shoes, what we divulge about ourselves and our habits has created a framework of information as it reveals hidden patterns in how we conduct our lives. The same can be true for our natural world as AI can help to reveal the patterns we haven't yet discovered.

We don't want to train VGG16 layers so we do *layer.trainable

This repository contains the trained model and dataset used for Unsupervised Adversarial Training (UAT) from the paper Are Labels Required for Improving Adversarial Robustness? Our model is available via TF-Hub. For example usage, refer to quick_eval_cifar.py. The preferred method of running this script is through run.sh, which will set up a virtual environment, install the dependendencies, and run the evaluation script, which will print the adversarial accuracy of the model. Note this file is very large, and requires 227 GB of disc space.