Search

State-of-the-art reinforcement learning (RL) algorithms suffer from high sample complexity, particularly in the sparse reward case. A popular strategy for mitigating this problem is to learn control policies by imitating a set of expert demonstrations. The drawback of such approaches is that an expert needs to produce demonstrations, which may be costly in practice. To address this shortcoming, we propose Probabilistic Planning for Demonstration Discovery (P2D2), a technique for automatically discovering demonstrations without access to an expert. We formulate discovering demonstrations as a search problem and leverage widely-used planning algorithms such as Rapidly-exploring Random Tree to find demonstration trajectories. These demonstrations are used to initialize a policy, then refined by a generic RL algorithm. We provide theoretical guarantees of P2D2 finding successful trajectories, as well as bounds for its sampling complexity. We experimentally demonstrate the method outperforms classic and intrinsic exploration RL techniques in a range of classic control and robotics tasks, requiring only a fraction of exploration samples and achieving better asymptotic performance., Comment: International Conference on Robotics and Automation (ICRA), 2021. arXiv admin note: substantial text overlap with arXiv:2001.06940

We propose a novel framework and algorithm for hierarchical planning based on the principle of delegation. This framework, the Markov Intent Process, features a collection of skills which are each specialised to perform a single task well. Skills are aware of their intended effects and are able to analyse planning goals to delegate planning to the best-suited skill. This principle dynamically creates a hierarchy of plans, in which each skill plans for sub-goals for which it is specialised. The proposed planning method features on-demand execution---skill policies are only evaluated when needed. Plans are only generated at the highest level, then expanded and optimised when the latest state information is available. The high-level plan retains the initial planning intent and previously computed skills, effectively reducing the computation needed to adapt to environmental changes. We show this planning approach is experimentally very competitive to classic planning and reinforcement learning techniques on a variety of domains, both in terms of solution length and planning time.

Intrinsic motivation enables reinforcement learning (RL) agents to explore when rewards are very sparse, where traditional exploration heuristics such as Boltzmann or e-greedy would typically fail. However, intrinsic exploration is generally handled in an ad-hoc manner, where exploration is not treated as a core objective of the learning process; this weak formulation leads to sub-optimal exploration performance. To overcome this problem, we propose a framework based on multi-objective RL where both exploration and exploitation are being optimized as separate objectives. This formulation brings the balance between exploration and exploitation at a policy level, resulting in advantages over traditional methods. This also allows for controlling exploration while learning, at no extra cost. Such strategies achieve a degree of control over agent exploration that was previously unattainable with classic or intrinsic rewards. We demonstrate scalability to continuous state-action spaces by presenting a method (EMU-Q) based on our framework, guiding exploration towards regions of higher value-function uncertainty. EMU-Q is experimentally shown to outperform classic exploration techniques and other intrinsic RL methods on a continuous control benchmark and on a robotic manipulator.

Balancing exploration and exploitation remains a key challenge in reinforcement learning (RL). State-of-the-art RL algorithms suffer from high sample complexity, particularly in the sparse reward case, where they can do no better than to explore in all directions until the first positive rewards are found. To mitigate this, we propose Rapidly Randomly-exploring Reinforcement Learning (R3L). We formulate exploration as a search problem and leverage widely-used planning algorithms such as Rapidly-exploring Random Tree (RRT) to find initial solutions. These solutions are used as demonstrations to initialize a policy, then refined by a generic RL algorithm, leading to faster and more stable convergence. We provide theoretical guarantees of R3L exploration finding successful solutions, as well as bounds for its sampling complexity. We experimentally demonstrate the method outperforms classic and intrinsic exploration techniques, requiring only a fraction of exploration samples and achieving better asymptotic performance.

Sampling-based planning is the predominant paradigm for motion planning in robotics. Most sampling-based planners use a global random sampling scheme to guarantee probabilistic completeness. However, most schemes are often inefficient as the samples drawn from the global proposal distribution, and do not exploit relevant local structures. Local sampling-based motion planners, on the other hand, take sequential decisions of random walks to samples valid trajectories in configuration space. However, current approaches do not adapt their strategies according to the success and failures of past samples. In this work, we introduce a local sampling-based motion planner with a Bayesian learning scheme for modelling an adaptive sampling proposal distribution. The proposal distribution is sequentially updated based on previous samples, consequently shaping it according to local obstacles and constraints in the configuration space. Thus, through learning from past observed outcomes, we maximise the likelihood of sampling in regions that have a higher probability to form trajectories within narrow passages. We provide the formulation of a sample-efficient distribution, along with theoretical foundation of sequentially updating this distribution. We demonstrate experimentally that by using a Bayesian proposal distribution, a solution is found faster, requiring fewer samples, and without any noticeable performance overhead., Comment: Accepted in IEEE Robotics and Automation Letters (RA-L)

We present a framework for learning to plan hierarchically in domains with unknown dynamics. We enhance planning performance by exploiting problem structure in several ways: (i) We simplify the search over plans by leveraging knowledge of skill objectives, (ii) Shorter plans are generated by enforcing aggressively hierarchical planning, (iii) We learn transition dynamics with sparse local models for better generalisation. Our framework decomposes transition dynamics into skill effects and success conditions, which allows fast planning by reasoning on effects, while learning conditions from interactions with the world. We propose a simple method for learning new abstract skills, using successful trajectories stemming from completing the goals of a curriculum. Learned skills are then refined to leverage other abstract skills and enhance subsequent planning. We show that both conditions and abstract skills can be learned simultaneously while planning, even in stochastic domains. Our method is validated in experiments of increasing complexity, with up to 2^100 states, showing superior planning to classic non-hierarchical planners or reinforcement learning methods. Applicability to real-world problems is demonstrated in a simulation-to-real transfer experiment on a robotic manipulator.

Bayesian Optimisation has gained much popularity lately, as a global optimisation technique for functions that are expensive to evaluate or unknown a priori. While classical BO focuses on where to gather an observation next, it does not take into account practical constraints for a robotic system such as where it is physically possible to gather samples from, nor the sequential nature of the problem while executing a trajectory. In field robotics and other real-life situations, physical and trajectory constraints are inherent problems. This paper addresses these issues by formulating Bayesian Optimisation for continuous trajectories within a Partially Observable Markov Decision Process (POMDP) framework. The resulting POMDP is solved using Monte-Carlo Tree Search (MCTS), which we adapt to using a reward function balancing exploration and exploitation. Experiments on monitoring a spatial phenomenon with a UAV illustrate how our BO-POMDP algorithm outperforms competing techniques., Comment: To appear in the 2017 IEEE International Conference on Robotics and Automation (ICRA)

Image instance retrieval is the problem of retrieving images from a database which contain the same object. Convolutional Neural Network (CNN) based descriptors are becoming the dominant approach for generating {\it global image descriptors} for the instance retrieval problem. One major drawback of CNN-based {\it global descriptors} is that uncompressed deep neural network models require hundreds of megabytes of storage making them inconvenient to deploy in mobile applications or in custom hardware. In this work, we study the problem of neural network model compression focusing on the image instance retrieval task. We study quantization, coding, pruning and weight sharing techniques for reducing model size for the instance retrieval problem. We provide extensive experimental results on the trade-off between retrieval performance and model size for different types of networks on several data sets providing the most comprehensive study on this topic. We compress models to the order of a few MBs: two orders of magnitude smaller than the uncompressed models while achieving negligible loss in retrieval performance., Comment: 10 pages, accepted by DCC 2017

The goal of this work is the computation of very compact binary hashes for image instance retrieval. Our approach has two novel contributions. The first one is Nested Invariance Pooling (NIP), a method inspired from i-theory, a mathematical theory for computing group invariant transformations with feed-forward neural networks. NIP is able to produce compact and well-performing descriptors with visual representations extracted from convolutional neural networks. We specifically incorporate scale, translation and rotation invariances but the scheme can be extended to any arbitrary sets of transformations. We also show that using moments of increasing order throughout nesting is important. The NIP descriptors are then hashed to the target code size (32-256 bits) with a Restricted Boltzmann Machine with a novel batch-level regularization scheme specifically designed for the purpose of hashing (RBMH). A thorough empirical evaluation with state-of-the-art shows that the results obtained both with the NIP descriptors and the NIP+RBMH hashes are consistently outstanding across a wide range of datasets., Comment: Image Instance Retrieval, CNN, Invariant Representation, Hashing, Unsupervised Learning, Regularization. arXiv admin note: text overlap with arXiv:1601.02093

Most image instance retrieval pipelines are based on comparison of vectors known as global image descriptors between a query image and the database images. Due to their success in large scale image classification, representations extracted from Convolutional Neural Networks (CNN) are quickly gaining ground on Fisher Vectors (FVs) as state-of-the-art global descriptors for image instance retrieval. While CNN-based descriptors are generally remarked for good retrieval performance at lower bitrates, they nevertheless present a number of drawbacks including the lack of robustness to common object transformations such as rotations compared with their interest point based FV counterparts. In this paper, we propose a method for computing invariant global descriptors from CNNs. Our method implements a recently proposed mathematical theory for invariance in a sensory cortex modeled as a feedforward neural network. The resulting global descriptors can be made invariant to multiple arbitrary transformation groups while retaining good discriminativeness. Based on a thorough empirical evaluation using several publicly available datasets, we show that our method is able to significantly and consistently improve retrieval results every time a new type of invariance is incorporated. We also show that our method which has few parameters is not prone to overfitting: improvements generalize well across datasets with different properties with regard to invariances. Finally, we show that our descriptors are able to compare favourably to other state-of-the-art compact descriptors in similar bitranges, exceeding the highest retrieval results reported in the literature on some datasets. A dedicated dimensionality reduction step --quantization or hashing-- may be able to further improve the competitiveness of the descriptors.

A typical image retrieval pipeline starts with the comparison of global descriptors from a large database to find a short list of candidate matches. A good image descriptor is key to the retrieval pipeline and should reconcile two contradictory requirements: providing recall rates as high as possible and being as compact as possible for fast matching. Following the recent successes of Deep Convolutional Neural Networks (DCNN) for large scale image classification, descriptors extracted from DCNNs are increasingly used in place of the traditional hand crafted descriptors such as Fisher Vectors (FV) with better retrieval performances. Nevertheless, the dimensionality of a typical DCNN descriptor --extracted either from the visual feature pyramid or the fully-connected layers-- remains quite high at several thousands of scalar values. In this paper, we propose Unsupervised Triplet Hashing (UTH), a fully unsupervised method to compute extremely compact binary hashes --in the 32-256 bits range-- from high-dimensional global descriptors. UTH consists of two successive deep learning steps. First, Stacked Restricted Boltzmann Machines (SRBM), a type of unsupervised deep neural nets, are used to learn binary embedding functions able to bring the descriptor size down to the desired bitrate. SRBMs are typically able to ensure a very high compression rate at the expense of loosing some desirable metric properties of the original DCNN descriptor space. Then, triplet networks, a rank learning scheme based on weight sharing nets is used to fine-tune the binary embedding functions to retain as much as possible of the useful metric properties of the original space. A thorough empirical evaluation conducted on multiple publicly available dataset using DCNN descriptors shows that our method is able to significantly outperform state-of-the-art unsupervised schemes in the target bit range.

With deep learning becoming the dominant approach in computer vision, the use of representations extracted from Convolutional Neural Nets (CNNs) is quickly gaining ground on Fisher Vectors (FVs) as favoured state-of-the-art global image descriptors for image instance retrieval. While the good performance of CNNs for image classification are unambiguously recognised, which of the two has the upper hand in the image retrieval context is not entirely clear yet. In this work, we propose a comprehensive study that systematically evaluates FVs and CNNs for image retrieval. The first part compares the performances of FVs and CNNs on multiple publicly available data sets. We investigate a number of details specific to each method. For FVs, we compare sparse descriptors based on interest point detectors with dense single-scale and multi-scale variants. For CNNs, we focus on understanding the impact of depth, architecture and training data on retrieval results. Our study shows that no descriptor is systematically better than the other and that performance gains can usually be obtained by using both types together. The second part of the study focuses on the impact of geometrical transformations such as rotations and scale changes. FVs based on interest point detectors are intrinsically resilient to such transformations while CNNs do not have a built-in mechanism to ensure such invariance. We show that performance of CNNs can quickly degrade in presence of rotations while they are far less affected by changes in scale. We then propose a number of ways to incorporate the required invariances in the CNN pipeline. Overall, our work is intended as a reference guide offering practically useful and simply implementable guidelines to anyone looking for state-of-the-art global descriptors best suited to their specific image instance retrieval problem., Comment: Deep Convolutional Neural Networks for instance retrieval, Fisher Vectors, instance retrieval

Compact keyframe-based video summaries are a popular way of generating viewership on video sharing platforms. Yet, creating relevant and compelling summaries for arbitrarily long videos with a small number of keyframes is a challenging task. We propose a comprehensive keyframe-based summarization framework combining deep convolutional neural networks and restricted Boltzmann machines. An original co-regularization scheme is used to discover meaningful subject-scene associations. The resulting multimodal representations are then used to select highly-relevant keyframes. A comprehensive user study is conducted comparing our proposed method to a variety of schemes, including the summarization currently in use by one of the most popular video sharing websites. The results show that our method consistently outperforms the baseline schemes for any given amount of keyframes both in terms of attractiveness and informativeness. The lead is even more significant for smaller summaries., Comment: Video summarization, deep convolutional neural networks, co-regularized restricted Boltzmann machines

This work focuses on representing very high-dimensional global image descriptors using very compact 64-1024 bit binary hashes for instance retrieval. We propose DeepHash: a hashing scheme based on deep networks. Key to making DeepHash work at extremely low bitrates are three important considerations -- regularization, depth and fine-tuning -- each requiring solutions specific to the hashing problem. In-depth evaluation shows that our scheme consistently outperforms state-of-the-art methods across all data sets for both Fisher Vectors and Deep Convolutional Neural Network features, by up to 20 percent over other schemes. The retrieval performance with 256-bit hashes is close to that of the uncompressed floating point features -- a remarkable 512 times compression.