**Note:** this version is tailored to @Smerity - though you can run your own! Trending arXiv may eventually be extended to multiple users ...

Imitation learning has been commonly applied to solve different tasks in isolation. This usually requires either careful feature engineering, or a significant number of samples. This is far from what we desire: ideally, robots should be able to learn from very few demonstrations of any given task, and instantly generalize to new situations of the same task, without requiring task-specific engineering. In this paper, we propose a meta-learning framework for achieving such capability, which we call one-shot imitation learning. Specifically, we consider the setting where there is a very large set of tasks, and each task has many instantiations. For example, a task could be to stack all blocks on a table into a single tower, another task could be to place all blocks on a table into two-block towers, etc. In each case, different instances of the task would consist of different sets of blocks with different initial states. At training time, our algorithm is presented with pairs of demonstrations for a subset of all tasks. A neural net is trained that takes as input one demonstration and the current state (which initially is the initial state of the other demonstration of the pair), and outputs an action with the goal that the resulting sequence of states and actions matches as closely as possible with the second demonstration. At test time, a demonstration of a single instance of a new task is presented, and the neural net is expected to perform well on new instances of this new task. The use of soft attention allows the model to generalize to conditions and tasks unseen in the training data. We anticipate that by training this model on a much greater variety of tasks and settings, we will obtain a general system that can turn any demonstrations into robust policies that can accomplish an overwhelming variety of tasks. Videos available at https://bit.ly/one-shot-imitation.

Captured tweets and retweets: 4

We present a conceptually simple, flexible, and general framework for object instance segmentation. Our approach efficiently detects objects in an image while simultaneously generating a high-quality segmentation mask for each instance. The method, called Mask R-CNN, extends Faster R-CNN by adding a branch for predicting an object mask in parallel with the existing branch for bounding box recognition. Mask R-CNN is simple to train and adds only a small overhead to Faster R-CNN, running at 5 fps. Moreover, Mask R-CNN is easy to generalize to other tasks, e.g., allowing us to estimate human poses in the same framework. We show top results in all three tracks of the COCO suite of challenges, including instance segmentation, bounding-box object detection, and person keypoint detection. Without tricks, Mask R-CNN outperforms all existing, single-model entries on every task, including the COCO 2016 challenge winners. We hope our simple and effective approach will serve as a solid baseline and help ease future research in instance-level recognition. Code will be made available.

Captured tweets and retweets: 102

We introduce the first goal-driven training for visual question answering and dialog agents. Specifically, we pose a cooperative 'image guessing' game between two agents -- Qbot and Abot -- who communicate in natural language dialog so that Qbot can select an unseen image from a lineup of images. We use deep reinforcement learning (RL) to learn the policies of these agents end-to-end -- from pixels to multi-agent multi-round dialog to game reward. We demonstrate two experimental results. First, as a 'sanity check' demonstration of pure RL (from scratch), we show results on a synthetic world, where the agents communicate in ungrounded vocabulary, i.e., symbols with no pre-specified meanings (X, Y, Z). We find that two bots invent their own communication protocol and start using certain symbols to ask/answer about certain visual attributes (shape/color/size). Thus, we demonstrate the emergence of grounded language and communication among 'visual' dialog agents with no human supervision at all. Second, we conduct large-scale real-image experiments on the VisDial dataset, where we pretrain with supervised dialog data and show that the RL 'fine-tuned' agents significantly outperform SL agents. Interestingly, the RL Qbot learns to ask questions that Abot is good at, ultimately resulting in more informative dialog and a better team.

Captured tweets and retweets: 2

Current Deep Learning approaches have been very successful using convolutional neural networks (CNN) trained on large graphical processing units (GPU)-based computers. Three limitations of this approach are: 1) they are based on a simple layered network topology, i.e., highly connected layers, without intra-layer connections; 2) the networks are manually configured to achieve optimal results, and 3) the implementation of neuron model is expensive in both cost and power. In this paper, we evaluate deep learning models using three different computing architectures to address these problems: quantum computing to train complex topologies, high performance computing (HPC) to automatically determine network topology, and neuromorphic computing for a low-power hardware implementation. We use the MNIST dataset for our experiment, due to input size limitations of current quantum computers. Our results show the feasibility of using the three architectures in tandem to address the above deep learning limitations. We show a quantum computer can find high quality values of intra-layer connections weights, in a tractable time as the complexity of the network increases; a high performance computer can find optimal layer-based topologies; and a neuromorphic computer can represent the complex topology and weights derived from the other architectures in low power memristive hardware.

Captured tweets and retweets: 1

While humans easily recognize relations between data from different domains without any supervision, learning to automatically discover them is in general very challenging and needs many ground-truth pairs that illustrate the relations. To avoid costly pairing, we address the task of discovering cross-domain relations given unpaired data. We propose a method based on generative adversarial networks that learns to discover relations between different domains (DiscoGAN). Using the discovered relations, our proposed network successfully transfers style from one domain to another while preserving key attributes such as orientation and face identity.

Captured tweets and retweets: 1

Many phenomena taking place in the solar photosphere are controlled by plasma motions. Although the line-of-sight component of the velocity can be estimated using the Doppler effect, we do not have direct spectroscopic access to the components that are perpendicular to the line-of-sight. These components are typically estimated using methods based on local correlation tracking. We have designed DeepVel, an end-to-end deep neural network that produces an estimation of the velocity at every single pixel and at every time step and at three different heights in the atmosphere from just two consecutive continuum images. We confront DeepVel with local correlation tracking, pointing out that they give very similar results in the time- and spatially-averaged cases. We use the network to study the evolution in height of the horizontal velocity field in fragmenting granules, supporting the buoyancy-braking mechanism for the formation of integranular lanes in these granules. We also show that DeepVel can capture very small vortices, so that we can potentially expand the scaling cascade of vortices to very small sizes and durations.

Captured tweets and retweets: 1

Despite their overwhelming capacity to overfit, deep learning architectures tend to generalize relatively well to unseen data, allowing them to be deployed in practice. However, explaining why this is the case is still an open area of research. One standing hypothesis that is gaining popularity, e.g. Hochreiter & Schmidhuber (1997); Keskar et al. (2017), is that the flatness of minima of the loss function found by stochastic gradient based methods results in good generalization. This paper argues that most notions of flatness are problematic for deep models and can not be directly applied to explain generalization. Specifically, when focusing on deep networks with rectifier units, we can exploit the particular geometry of parameter space induced by the inherent symmetries that these architectures exhibit to build equivalent models corresponding to arbitrarily sharper minima. Furthermore, if we allow to reparametrize a function, the geometry of its parameters can change drastically without affecting its generalization properties.

Captured tweets and retweets: 2

Increasing evidence suggests that a growing amount of social media content is generated by autonomous entities known as social bots. In this work we present a framework to detect such entities on Twitter. We leverage more than a thousand features extracted from public data and meta-data about users: friends, tweet content and sentiment, network patterns, and activity time series. We benchmark the classification framework by using a publicly available dataset of Twitter bots. This training data is enriched by a manually annotated collection of active Twitter users that include both humans and bots of varying sophistication. Our models yield high accuracy and agreement with each other and can detect bots of different nature. Our estimates suggest that between 9% and 15% of active Twitter accounts are bots. Characterizing ties among accounts, we observe that simple bots tend to interact with bots that exhibit more human-like behaviors. Analysis of content flows reveals retweet and mention strategies adopted by bots to interact with different target groups. Using clustering analysis, we characterize several subclasses of accounts, including spammers, self promoters, and accounts that post content from connected applications.

Captured tweets and retweets: 8

Traditional image and video compression algorithms rely on hand-crafted encoder/decoder pairs (codecs) that lack adaptability and are agnostic to the data being compressed. Here we describe the concept of generative compression, the compression of data using generative models, and show its potential to produce more accurate and visually pleasing reconstructions at much deeper compression levels for both image and video data. We also demonstrate that generative compression is orders-of-magnitude more resilient to bit error rates (e.g. from noisy wireless channels) than traditional variable-length entropy coding schemes.

Captured tweets and retweets: 2

Bellemare et al. (2016) introduced the notion of a pseudo-count to generalize count-based exploration to non-tabular reinforcement learning. This pseudo-count is derived from a density model which effectively replaces the count table used in the tabular setting. Using an exploration bonus based on this pseudo-count and a mixed Monte Carlo update applied to a DQN agent was sufficient to achieve state-of-the-art on the Atari 2600 game Montezuma's Revenge. In this paper we consider two questions left open by their work: First, how important is the quality of the density model for exploration? Second, what role does the Monte Carlo update play in exploration? We answer the first question by demonstrating the use of PixelCNN, an advanced neural density model for images, to supply a pseudo-count. In particular, we examine the intrinsic difficulties in adapting Bellemare et al's approach when assumptions about the model are violated. The result is a more practical and general algorithm requiring no special apparatus. We combine PixelCNN pseudo-counts with different agent architectures to dramatically improve the state of the art on several hard Atari games. One surprising finding is that the mixed Monte Carlo update is a powerful facilitator of exploration in the sparsest of settings, including Montezuma's Revenge.

Captured tweets and retweets: 1

Most of the existing image-to-image translation frameworks---mapping an image in one domain to a corresponding image in another---are based on supervised learning, i.e., pairs of corresponding images in two domains are required for learning the translation function. This largely limits their applications, because capturing corresponding images in two different domains is often a difficult task. To address the issue, we propose the UNsupervised Image-to-image Translation (UNIT) framework, which is based on variational autoencoders and generative adversarial networks. The proposed framework can learn the translation function without any corresponding images in two domains. We enable this learning capability by combining a weight-sharing constraint and an adversarial training objective. Through visualization results from various unsupervised image translation tasks, we verify the effectiveness of the proposed framework. An ablation study further reveals the critical design choices. Moreover, we apply the UNIT framework to the unsupervised domain adaptation task and achieve better results than competing algorithms do in benchmark datasets.

Captured tweets and retweets: 2

Sophisticated gated recurrent neural network architectures like LSTMs and GRUs have been shown to be highly effective in a myriad of applications. We develop an un-gated unit, the statistical recurrent unit (SRU), that is able to learn long term dependencies in data by only keeping moving averages of statistics. The SRU's architecture is simple, un-gated, and contains a comparable number of parameters to LSTMs; yet, SRUs perform favorably to more sophisticated LSTM and GRU alternatives, often outperforming one or both in various tasks. We show the efficacy of SRUs as compared to LSTMs and GRUs in an unbiased manner by optimizing respective architectures' hyperparameters in a Bayesian optimization scheme for both synthetic and real-world tasks.

Captured tweets and retweets: 1

In this paper, we propose gcForest, a decision tree ensemble approach with performance highly competitive to deep neural networks. In contrast to deep neural networks which require great effort in hyper-parameter tuning, gcForest is much easier to train. Actually, even when gcForest is applied to different data from different domains, excellent performance can be achieved by almost same settings of hyper-parameters. The training process of gcForest is efficient and scalable. In our experiments its training time running on a PC is comparable to that of deep neural networks running with GPU facilities, and the efficiency advantage may be more apparent because gcForest is naturally apt to parallel implementation. Furthermore, in contrast to deep neural networks which require large-scale training data, gcForest can work well even when there are only small-scale training data. Moreover, as a tree-based approach, gcForest should be easier for theoretical analysis than deep neural networks.

Captured tweets and retweets: 10

Deep Residual Networks have reached the state of the art in many image processing tasks such image classification. However, the cost for a gain in accuracy in terms of depth and memory is prohibitive as it requires a higher number of residual blocks, up to double the initial value. To tackle this problem, we propose in this paper a way to reduce the redundant information of the networks. We share the weights of convolutional layers between residual blocks operating at the same spatial scale. The signal flows multiple times in the same convolutional layer. The resulting architecture, called ShaResNet, contains block specific layers and shared layers. These ShaResNet are trained exactly in the same fashion as the commonly used residual networks. We show, on the one hand, that they are almost as efficient as their sequential counterparts while involving less parameters, and on the other hand that they are more efficient than a residual network with the same number of parameters. For example, a 152-layer-deep residual network can be reduced to 106 convolutional layers, i.e. a parameter gain of 39\%, while loosing less than 0.2\% accuracy on ImageNet.

Captured tweets and retweets: 2

Similarity search finds application in specialized database systems handling complex data such as images or videos, which are typically represented by high-dimensional features and require specific indexing structures. This paper tackles the problem of better utilizing GPUs for this task. While GPUs excel at data-parallel tasks, prior approaches are bottlenecked by algorithms that expose less parallelism, such as k-min selection, or make poor use of the memory hierarchy. We propose a design for k-selection that operates at up to 55% of theoretical peak performance, enabling a nearest neighbor implementation that is 8.5x faster than prior GPU state of the art. We apply it in different similarity search scenarios, by proposing optimized design for brute-force, approximate and compressed-domain search based on product quantization. In all these setups, we outperform the state of the art by large margins. Our implementation enables the construction of a high accuracy k-NN graph on 95 million images from the Yfcc100M dataset in 35 minutes, and of a graph connecting 1 billion vectors in less than 12 hours on 4 Maxwell Titan X GPUs. We have open-sourced our approach for the sake of comparison and reproducibility.

Captured tweets and retweets: 1

We introduce AI rationalization, an approach for generating explanations of autonomous system behavior as if a human had done the behavior. We describe a rationalization technique that uses neural machine translation to translate internal state-action representations of the autonomous agent into natural language. We evaluate our technique in the Frogger game environment. The natural language is collected from human players thinking out loud as they play the game. We motivate the use of rationalization as an approach to explanation generation, show the results of experiments on the accuracy of our rationalization technique, and describe future research agenda.

Captured tweets and retweets: 2

The United States spends more than $1B each year on the American Community Survey (ACS), a labor-intensive door-to-door study that measures statistics relating to race, gender, education, occupation, unemployment, and other demographic factors. Although a comprehensive source of data, the lag between demographic changes and their appearance in the ACS can exceed half a decade. As digital imagery becomes ubiquitous and machine vision techniques improve, automated data analysis may provide a cheaper and faster alternative. Here, we present a method that determines socioeconomic trends from 50 million images of street scenes, gathered in 200 American cities by Google Street View cars. Using deep learning-based computer vision techniques, we determined the make, model, and year of all motor vehicles encountered in particular neighborhoods. Data from this census of motor vehicles, which enumerated 22M automobiles in total (8% of all automobiles in the US), was used to accurately estimate income, race, education, and voting patterns, with single-precinct resolution. (The average US precinct contains approximately 1000 people.) The resulting associations are surprisingly simple and powerful. For instance, if the number of sedans encountered during a 15-minute drive through a city is higher than the number of pickup trucks, the city is likely to vote for a Democrat during the next Presidential election (88% chance); otherwise, it is likely to vote Republican (82%). Our results suggest that automated systems for monitoring demographic trends may effectively complement labor-intensive approaches, with the potential to detect trends with fine spatial resolution, in close to real time.

Captured tweets and retweets: 2

We explore design principles for general pixel-level prediction problems, from low-level edge detection to mid-level surface normal estimation to high-level semantic segmentation. Convolutional predictors, such as the fully-convolutional network (FCN), have achieved remarkable success by exploiting the spatial redundancy of neighboring pixels through convolutional processing. Though computationally efficient, we point out that such approaches are not statistically efficient during learning precisely because spatial redundancy limits the information learned from neighboring pixels. We demonstrate that stratified sampling of pixels allows one to (1) add diversity during batch updates, speeding up learning; (2) explore complex nonlinear predictors, improving accuracy; and (3) efficiently train state-of-the-art models tabula rasa (i.e., "from scratch") for diverse pixel-labeling tasks. Our single architecture produces state-of-the-art results for semantic segmentation on PASCAL-Context dataset, surface normal estimation on NYUDv2 depth dataset, and edge detection on BSDS.

Captured tweets and retweets: 15

We extend Fano's inequality, which controls the average probability of (disjoint) events in terms of the average of some Kullback-Leibler divergences, to work with arbitrary [0,1]-valued random variables. Our simple two-step methodology is general enough to cover the case of an arbitrary (possibly continuously infinite) family of distributions as well as [0,1]-valued random variables not necessarily summing up to 1. Several novel applications are provided, in which the consideration of random variables is particularly handy. The most important applications deal with the problem of Bayesian posterior concentration (minimax or distribution-dependent) rates and with a lower bound on the regret in non-stochastic sequential learning. We also improve in passing some earlier fundamental results: in particular, we provide a simple and enlightening proof of the refined Pinsker's inequality of Ordentlich and Weinberger and derive a sharper Bretagnolle-Huber inequality.

Captured tweets and retweets: 2