Autoregressive sequence Generation models have achieved state-of-the-art performance in areas like machine translation and image captioning. These models are autoregressive in that they generate each word by conditioning on previously generated words, which leads to heavy latency during inference. Recently, non-autoregressive decoding has been proposed in machine translation to speed up the inference time by generating all words in parallel. Typically, these models use the word-level cross-entropy loss to optimize each word independently. However, such a learning process fails to consider the sentence-level consistency, thus resulting in inferior generation quality of these non-autoregressive models. In this paper, we propose a simple and efficient model for Non-Autoregressive sequence Generation (NAG) with a novel training paradigm: Counterfactuals-critical Multi-Agent Learning (CMAL). CMAL formulates NAG as a multi-agent reinforcement learning system where element positions in the target sequence are viewed as agents that learn to cooperatively maximize a sentence-level reward. On MSCOCO image captioning benchmark, our NAG method achieves a performance comparable to state-of-the-art autoregressive models, while brings 13.9x decoding speedup. On WMT14 EN-DE machine translation dataset, our method outperforms cross-entropy trained baseline by 6.0 BLEU points while achieves the greatest decoding speedup of 17.46x.

Many complex processes can be viewed as dynamical systems of interacting agents. In many cases, only the state sequences of individual agents are observed, while the interacting relations and the dynamical rules are unknown. The neural relational inference (NRI) model adopts graph neural networks that pass messages over a latent graph to jointly learn the relations and the dynamics based on the observed data. However, NRI infers the relations independently and suffers from error accumulation in multi-step prediction at dynamics learning procedure. Besides, relation reconstruction without prior knowledge becomes more difficult in more complex systems. This paper introduces efficient message passing mechanisms to the graph neural networks with structural prior knowledge to address these problems. A relation interaction mechanism is proposed to capture the coexistence of all relations, and a spatio-temporal message passing mechanism is proposed to use historical information to alleviate error accumulation. Additionally, the structural prior knowledge, symmetry as a special case, is introduced for better relation prediction in more complex systems. The experimental results on simulated physics systems show that the proposed method outperforms existing state-of-the-art methods.

Deep reinforcement learning (DRL) is applied in safety-critical domains such as robotics and autonomous driving. It achieves superhuman abilities in many tasks, however whether DRL agents can be shown to act safely is an open problem. Atari games are a simple yet challenging exemplar for evaluating the safety of DRL agents and feature a diverse portfolio of game mechanics. The safety of neural agents has been studied before using methods that either require a model of the system dynamics or an abstraction; unfortunately, these are unsuitable to Atari games because their low-level dynamics are complex and hidden inside their emulator. We present the first exact method for analysing and ensuring the safety of DRL agents for Atari games. Our method only requires access to the emulator. First, we give a set of 43 properties that characterise "safe behaviour" for 30 games. Second, we develop a method for exploring all traces induced by an agent and a game and consider a variety of sources of game non-determinism. We observe that the best available DRL agents reliably satisfy only very few properties; several critical properties are violated by all agents. Finally, we propose a countermeasure that combines a bounded explicit-state exploration with shielding. We demonstrate that our method improves the safety of all agents over multiple properties.

Multi-agent interacting systems are prevalent in the world, from purely physical systems to complicated social dynamic systems. The interactions between entities / components can give rise to very complex behavior patterns at the level of both individuals and the multi-agent system as a whole. Since usually only the trajectories of individual entities are observed without any knowledge of the underlying interaction patterns, and there are usually multiple possible modalities for each agent with uncertainty, it is challenging to model their dynamics and forecast their future behaviors. We introduce a generic trajectory forecasting framework (named EvolveGraph) with explicit relational structure recognition and prediction via latent interaction graphs among multiple heterogeneous, interactive agents. Considering the uncertainty of future behaviors, the model is designed to provide multi-modal prediction hypotheses.

Autonomous trading robots have been studied in ar-tificial intelligence area for quite some time. Many AI techniqueshave been tested in finance field including recent approaches likeconvolutional neural networks and deep reinforcement learning.There are many reported cases, where the developers are suc-cessful in creating robots with great performance when executingwith historical price series, so called backtesting. However, whenthese robots are used in real markets or data not used intheir training or evaluation frequently they present very poorperformance in terms of risks and return. In this paper, wediscussed some fundamental aspects of modelling autonomoustraders and the complex environment that is the financialworld. Furthermore, we presented a framework that helps thedevelopment and testing of autonomous traders. It may also beused in real or simulated operation in financial markets. Finally,we discussed some open problems in the area and pointed outsome interesting technologies that may contribute to advancein such task. We believe that mt5b3 may also contribute todevelopment of new autonomous traders.

Wind farms are a crucial driver toward the generation of ecological and renewable energy. Due to their rapid increase in capacity, contemporary wind farms need to adhere to strict constraints on power output to ensure stability of the electricity grid. Specifically, a wind farm controller is required to match the farm's power production with a power demand imposed by the grid operator. This is a non-trivial optimization problem, as complex dependencies exist between the wind turbines. State-of-the-art wind farm control typically relies on physics-based heuristics that fail to capture the full load spectrum that defines a turbine's health status. When this is not taken into account, the long-term viability of the farm's turbines is put at risk. Given the complex dependencies that determine a turbine's lifetime, learning a flexible and optimal control strategy requires a data-driven approach. However, as wind farms are large-scale multi-agent systems, optimizing control strategies over the full joint action space is intractable. We propose a new learning method for wind farm control that leverages the sparse wind farm structure to factorize the optimization problem. Using a Bayesian approach, based on multi-agent Thompson sampling, we explore the factored joint action space for configurations that match the demand, while considering the lifetime of turbines. We apply our method to a grid-like wind farm layout, and evaluate configurations using a state-of-the-art wind flow simulator. Our results are competitive with a physics-based heuristic approach in terms of demand error, while, contrary to the heuristic, our method prolongs the lifetime of high-risk turbines.

For some scientific questions, empirical data are essential to develop reliable simulation models. These data usually come from different sources with diverse and heterogeneous formats. The design of complex data-driven models is often shaped by the structure of the data available in research projects. Hence, applying such models to other case studies requires either to get similar data or to transform new data to fit the model inputs. It is the case of agent-based models (ABMs) that use advanced data structures such as Geographic Information Systems data. We faced this problem in the LittoSIM-GEN project when generalizing our participatory flooding model (LittoSIM) to new territories. From this experience, we provide a mapping approach to structure, describe, and automatize the integration of geospatial data into ABMs.

Cooperative multi-agent reinforcement learning (MARL) has achieved significant results, most notably by leveraging the representation learning abilities of deep neural networks. However, large centralized approaches quickly become infeasible as the number of agents scale, and fully decentralized approaches can miss important opportunities for information sharing and coordination. Furthermore, not all agents are equal - in some cases, individual agents may not even have the ability to send communication to other agents or explicitly model other agents. This paper considers the case where there is a single, powerful, central agent that can observe the entire observation space, and there are multiple, low powered, local agents that can only receive local observations and cannot communicate with each other. The job of the central agent is to learn what message to send to different local agents, based on the global observations, not by centrally solving the entire problem and sending action commands, but by determining what additional information an individual agent should receive so that it can make a better decision. After explaining our MARL algorithm, hammer, and where it would be most applicable, we implement it in the cooperative navigation and multi-agent walker domains. Empirical results show that 1) learned communication does indeed improve system performance, 2) results generalize to multiple numbers of agents, and 3) results generalize to different reward structures.

Recent years saw a plethora of work on explaining complex intelligent agents. One example is the development of several algorithms that generate saliency maps which show how much each pixel attributed to the agents' decision. However, most evaluations of such saliency maps focus on image classification tasks. As far as we know, there is no work which thoroughly compares different saliency maps for Deep Reinforcement Learning agents. This paper compares four perturbation-based approaches to create saliency maps for Deep Reinforcement Learning agents trained on four different Atari 2600 games. All four approaches work by perturbing parts of the input and measuring how much this affects the agent's output. The approaches are compared using three computational metrics: dependence on the learned parameters of the agent (sanity checks), faithfulness to the agent's reasoning (input degradation), and run-time.

Modeling mixed-traffic motion and interactions is crucial to assess safety, efficiency, and feasibility of future urban areas. The lack of traffic regulations, diverse transport modes, and the dynamic nature of mixed-traffic zones like shared spaces make realistic modeling of such environments challenging. This paper focuses on the generalizability of the motion model, i.e., its ability to generate realistic behavior in different environmental settings, an aspect which is lacking in existing works. Specifically, our first contribution is a novel and systematic process of formulating general motion models and application of this process is to extend our Game-Theoretic Social Force Model (GSFM) towards a general model for generating a large variety of motion behaviors of pedestrians and cars from different shared spaces. Our second contribution is to consider different motion patterns of pedestrians by calibrating motion-related features of individual pedestrian and clustering them into groups. We analyze two clustering approaches. The calibration and evaluation of our model are performed on three different shared space data sets. The results indicate that our model can realistically simulate a wide range of motion behaviors and interaction scenarios, and that adding different motion patterns of pedestrians into our model improves its performance.