Training intelligent agents through reinforcement learning is a notoriously unstable procedure. Massive parallelization on GPUs and distributed systems has been exploited to generate a large amount of training experiences and consequently reduce instabilities, but the success of training remains strongly influenced by the choice of the hyperparameters. To overcome this issue, we introduce HyperTrick, a new metaoptimization algorithm, and show its effective application to tune hyperparameters in the case of deep reinforcement learning, while learning to play different Atari games on a distributed system. Our analysis provides evidence of the interaction between the identification of the optimal hyperparameters and the learned policy, that is typical of the case of metaoptimization for deep reinforcement learning. When compared with state-of-the-art metaoptimization algorithms, HyperTrick is characterized by a simpler implementation and it allows learning similar policies, while making a more effective use of the computational resources in a distributed system.
Scene motion, multiple reflections, and sensor noise introduce artifacts in the depth reconstruction performed by time-of-flight cameras. We propose a two-stage, deep-learning approach to address all of these sources of artifacts simultaneously. We also introduce FLAT, a synthetic dataset of 2000 ToF measurements that capture all of these nonidealities, and allows to simulate different camera hardware. Using the Kinect 2 camera as a baseline, we show improved reconstruction errors over state-of-the-art methods, on both simulated and real data.
Neural networks are becoming central in several areas of computer vision and image processing and different architectures have been proposed to solve specific problems. The impact of the loss layer of neural networks, however, has not received much attention in the context of image processing: the default and virtually only choice is L2. In this paper, we bring attention to alternative choices for image restoration. In particular, we show the importance of perceptually-motivated losses when the resulting image is to be evaluated by a human observer. We compare the performance of several losses, and propose a novel, differentiable error function. We show that the quality of the results improves significantly with better loss functions, even when the network architecture is left unchanged.
The non-stationary nature of image characteristics calls for adaptive processing, based on the local image content. We propose a simple and flexible method to learn local tuning of parameters in adaptive image processing: we extract simple local features from an image and learn the relation between these features and the optimal filtering parameters. Learning is performed by optimizing a user defined cost function (any image quality metric) on a training set. We apply our method to three classical problems (denoising, demosaicing and deblurring) and we show the effectiveness of the learned parameter modulation strategies. We also show that these strategies are consistent with theoretical results from the literature.
In the context of deep learning for robotics, we show effective method of training a real robot to grasp a tiny sphere (1.37cm of diameter), with an original combination of system design choices. We decompose the end-to-end system into a vision module and a closed-loop controller module. The two modules use target object segmentation as their common interface. The vision module extracts information from the robot end-effector camera, in the form of a binary segmentation mask of the target. We train it to achieve effective domain transfer by composing real background images with simulated images of the target. The controller module takes as input the binary segmentation mask, and thus is agnostic to visual discrepancies between simulated and real environments. We train our closed-loop controller in simulation using imitation learning and show it is robust with respect to discrepancies between the dynamic model of the simulated and real robot: when combined with eye-in-hand observations, we achieve a 90% success rate in grasping a tiny sphere with a real robot. The controller can generalize to unseen scenarios where the target is moving and even learns to recover from failures.
To denoise a reference patch, the Non-Local-Means denoising filter processes a set of neighbor patches. Few Nearest Neighbors (NN) are used to limit the computational burden of the algorithm. Here here we show analytically that the NN approach introduces a bias in the denoised patch, and we propose a different neighbors' collection criterion, named Statistical NN (SNN), to alleviate this issue. Our approach outperforms the traditional one in case of both white and colored noise: fewer SNNs generate images of higher quality, at a lower computational cost.
We introduce a hybrid CPU/GPU version of the Asynchronous Advantage Actor-Critic (A3C) algorithm, currently the state-of-the-art method in reinforcement learning for various gaming tasks. We analyze its computational traits and concentrate on aspects critical to leveraging the GPU's computational power. We introduce a system of queues and a dynamic scheduling strategy, potentially helpful for other asynchronous algorithms as well. Our hybrid CPU/GPU version of A3C, based on TensorFlow, achieves a significant speed up compared to a CPU implementation; we make it publicly available to other researchers at https://github.com/NVlabs/GA3C .