The last decade has seen a significant increase of interest in deep learning research, with many public successes that have demonstrated its potential. As such, these systems are now being incorporated into commercial products. With this comes an additional challenge: how can we build AI systems that solve tasks where there is not a crisp, well-defined specification? While multiple solutions have been proposed, in this competition we focus on one in particular: learning from human feedback. Rather than training AI systems using a predefined reward function or using a labeled dataset with a predefined set of categories, we instead train the AI system using a learning signal derived from some form of human feedback, which can evolve over time as the understanding of the task changes, or as the capabilities of the AI system improve. The MineRL BASALT competition aims to spur forward research on this important class of techniques. We design a suite of four tasks in Minecraft for which we expect it will be hard to write down hardcoded reward functions. These tasks are defined by a paragraph of natural language: for example, "create a waterfall and take a scenic picture of it", with additional clarifying details. Participants must train a separate agent for each task, using any method they want. Agents are then evaluated by humans who have read the task description. To help participants get started, we provide a dataset of human demonstrations on each of the four tasks, as well as an imitation learning baseline that leverages these demonstrations. Our hope is that this competition will improve our ability to build AI systems that do what their designers intend them to do, even when the intent cannot be easily formalized. Besides allowing AI to solve more tasks, this can also enable more effective regulation of AI systems, as well as making progress on the value alignment problem.
An important challenge in reinforcement learning is training agents that can solve a wide variety of tasks. If tasks depend on each other (e.g. needing to learn to walk before learning to run), curriculum learning can speed up learning by focusing on the next best task to learn. We explore curriculum learning in a complex, visual domain with many hard exploration challenges: Minecraft. We find that learning progress (defined as a change in success probability of a task) is a reliable measure of learnability for automatically constructing an effective curriculum. We introduce a learning-progress based curriculum and test it on a complex reinforcement learning problem (called "Simon Says") where an agent is instructed to obtain a desired goal item. Many of the required skills depend on each other. Experiments demonstrate that: (1) a within-episode exploration bonus for obtaining new items improves performance, (2) dynamically adjusting this bonus across training such that it only applies to items the agent cannot reliably obtain yet further increases performance, (3) the learning-progress based curriculum elegantly follows the learning curve of the agent, and (4) when the learning-progress based curriculum is combined with the dynamic exploration bonus it learns much more efficiently and obtains far higher performance than uniform baselines. These results suggest that combining intra-episode and across-training exploration bonuses with learning progress creates a promising method for automated curriculum generation, which may substantially increase our ability to train more capable, generally intelligent agents.
This work introduces World-GAN, the first method to perform data-driven Procedural Content Generation via Machine Learning in Minecraft from a single example. Based on a 3D Generative Adversarial Network (GAN) architecture, we are able to create arbitrarily sized world snippets from a given sample. We evaluate our approach on creations from the community as well as structures generated with the Minecraft World Generator. Our method is motivated by the dense representations used in Natural Language Processing (NLP) introduced with word2vec [1]. The proposed block2vec representations make World-GAN independent from the number of different blocks, which can vary a lot in Minecraft, and enable the generation of larger levels. Finally, we demonstrate that changing this new representation space allows us to change the generated style of an already trained generator. World-GAN enables its users to generate Minecraft worlds based on parts of their creations.
Task environments developed in Minecraft are becoming increasingly popular for artificial intelligence (AI) research. However, most of these are currently constructed manually, thus failing to take advantage of procedural content generation (PCG), a capability unique to virtual task environments. In this paper, we present mcg, an open-source library to facilitate implementing PCG algorithms for voxel-based environments such as Minecraft. The library is designed with human-machine teaming research in mind, and thus takes a 'top-down' approach to generation, simultaneously generating low and high level machine-readable representations that are suitable for empirical research. These can be consumed by downstream AI applications that consider human spatial cognition. The benefits of this approach include rapid, scalable, and efficient development of virtual environments, the ability to control the statistics of the environment at a semantic level, and the ability to generate novel environments in response to player actions in real time.
We present GANcraft, an unsupervised neural rendering framework for generating photorealistic images of large 3D block worlds such as those created in Minecraft. Our method takes a semantic block world as input, where each block is assigned a semantic label such as dirt, grass, or water. We represent the world as a continuous volumetric function and train our model to render view-consistent photorealistic images for a user-controlled camera. In the absence of paired ground truth real images for the block world, we devise a training technique based on pseudo-ground truth and adversarial training. This stands in contrast to prior work on neural rendering for view synthesis, which requires ground truth images to estimate scene geometry and view-dependent appearance. In addition to camera trajectory, GANcraft allows user control over both scene semantics and output style. Experimental results with comparison to strong baselines show the effectiveness of GANcraft on this novel task of photorealistic 3D block world synthesis. The project website is available at https://nvlabs.github.io/GANcraft/ .
When developing AI systems that interact with humans, it is essential to design both a system that can understand humans, and a system that humans can understand. Most deep network based agent-modeling approaches are 1) not interpretable and 2) only model external behavior, ignoring internal mental states, which potentially limits their capability for assistance, interventions, discovering false beliefs, etc. To this end, we develop an interpretable modular neural framework for modeling the intentions of other observed entities. We demonstrate the efficacy of our approach with experiments on data from human participants on a search and rescue task in Minecraft, and show that incorporating interpretability can significantly increase predictive performance under the right conditions.
This article outlines what we learned from the first year of the AI Settlement Generation Competition in Minecraft, a competition about producing AI programs that can generate interesting settlements in Minecraft for an unseen map. This challenge seeks to focus research into adaptive and holistic procedural content generation. Generating Minecraft towns and villages given existing maps is a suitable task for this, as it requires the generated content to be adaptive, functional, evocative and aesthetic at the same time. Here, we present the results from the first iteration of the competition. We discuss the evaluation methodology, present the different technical approaches by the competitors, and outline the open problems.
Neural Cellular Automata (NCAs) have been proven effective in simulating morphogenetic processes, the continuous construction of complex structures from very few starting cells. Recent developments in NCAs lie in the 2D domain, namely reconstructing target images from a single pixel or infinitely growing 2D textures. In this work, we propose an extension of NCAs to 3D, utilizing 3D convolutions in the proposed neural network architecture. Minecraft is selected as the environment for our automaton since it allows the generation of both static structures and moving machines. We show that despite their simplicity, NCAs are capable of growing complex entities such as castles, apartment blocks, and trees, some of which are composed of over 3,000 blocks. Additionally, when trained for regeneration, the system is able to regrow parts of simple functional machines, significantly expanding the capabilities of simulated morphogenetic systems. The code for the experiment in this paper can be found at: https://github.com/real-itu/3d-artefacts-nca.
In the univariate setting, using the kernel spectral representation is an appealing approach for generating stationary covariance functions. However, performing the same task for multiple-output Gaussian processes is substantially more challenging. We demonstrate that current approaches to modelling cross-covariances with a spectral mixture kernel possess a critical blind spot. For a given pair of processes, the cross-covariance is not reproducible across the full range of permitted correlations, aside from the special case where their spectral densities are of identical shape. We present a solution to this issue by replacing the conventional Gaussian components of a spectral mixture with block components of finite bandwidth (i.e. rectangular step functions). The proposed family of kernel represents the first multi-output generalisation of the spectral mixture kernel that can approximate any stationary multi-output kernel to arbitrary precision.
Learning to flexibly follow task instructions in dynamic environments poses interesting challenges for reinforcement learning agents. We focus here on the problem of learning control flow that deviates from a strict step-by-step execution of instructions -- that is, control flow that may skip forward over parts of the instructions or return backward to previously completed or skipped steps. Demand for such flexible control arises in two fundamental ways: explicitly when control is specified in the instructions themselves (such as conditional branching and looping) and implicitly when stochastic environment dynamics require re-completion of instructions whose effects have been perturbed, or opportunistic skipping of instructions whose effects are already present. We formulate an attention-based architecture that meets these challenges by learning, from task reward only, to flexibly attend to and condition behavior on an internal encoding of the instructions. We test the architecture's ability to learn both explicit and implicit control in two illustrative domains -- one inspired by Minecraft and the other by StarCraft -- and show that the architecture exhibits zero-shot generalization to novel instructions of length greater than those in a training set, at a performance level unmatched by two baseline recurrent architectures and one ablation architecture.
A key challenge for reinforcement learning is solving long-horizon planning problems. Recent work has leveraged programs to guide reinforcement learning in these settings. However, these approaches impose a high manual burden on the user since they must provide a guiding program for every new task. Partially observed environments further complicate the programming task because the program must implement a strategy that correctly, and ideally optimally, handles every possible configuration of the hidden regions of the environment. We propose a new approach, model predictive program synthesis (MPPS), that uses program synthesis to automatically generate the guiding programs. It trains a generative model to predict the unobserved portions of the world, and then synthesizes a program based on samples from this model in a way that is robust to its uncertainty. In our experiments, we show that our approach significantly outperforms non-program-guided approaches on a set of challenging benchmarks, including a 2D Minecraft-inspired environment where the agent must complete a complex sequence of subtasks to achieve its goal, and achieves a similar performance as using handcrafted programs to guide the agent. Our results demonstrate that our approach can obtain the benefits of program-guided reinforcement learning without requiring the user to provide a new guiding program for every new task.
Deep learning has enabled algorithms to generate realistic images. However, accurately predicting long video sequences requires understanding long-term dependencies and remains an open challenge. While existing video prediction models succeed at generating sharp images, they tend to fail at accurately predicting far into the future. We introduce the Clockwork VAE (CW-VAE), a video prediction model that leverages a hierarchy of latent sequences, where higher levels tick at slower intervals. We demonstrate the benefits of both hierarchical latents and temporal abstraction on 4 diverse video prediction datasets with sequences of up to 1000 frames, where CW-VAE outperforms top video prediction models. Additionally, we propose a Minecraft benchmark for long-term video prediction. We conduct several experiments to gain insights into CW-VAE and confirm that slower levels learn to represent objects that change more slowly in the video, and faster levels learn to represent faster objects.
Although deep reinforcement learning has led to breakthroughs in many difficult domains, these successes have required an ever-increasing number of samples, affording only a shrinking segment of the AI community access to their development. Resolution of these limitations requires new, sample-efficient methods. To facilitate research in this direction, we propose this second iteration of the MineRL Competition. The primary goal of the competition is to foster the development of algorithms which can efficiently leverage human demonstrations to drastically reduce the number of samples needed to solve complex, hierarchical, and sparse environments. To that end, participants compete under a limited environment sample-complexity budget to develop systems which solve the MineRL ObtainDiamond task in Minecraft, a sequential decision making environment requiring long-term planning, hierarchical control, and efficient exploration methods. The competition is structured into two rounds in which competitors are provided several paired versions of the dataset and environment with different game textures and shaders. At the end of each round, competitors submit containerized versions of their learning algorithms to the AIcrowd platform where they are trained from scratch on a hold-out dataset-environment pair for a total of 4-days on a pre-specified hardware platform. In this follow-up iteration to the NeurIPS 2019 MineRL Competition, we implement new features to expand the scale and reach of the competition. In response to the feedback of the previous participants, we introduce a second minor track focusing on solutions without access to environment interactions of any kind except during test-time. Further we aim to prompt domain agnostic submissions by implementing several novel competition mechanics including action-space randomization and desemantization of observations and actions.
Reinforcement learning has enabled agents to solve challenging tasks in unknown environments. However, manually crafting reward functions can be time consuming, expensive, and error prone to human error. Competing objectives have been proposed for agents to learn without external supervision, but it has been unclear how well they reflect task rewards or human behavior. To accelerate the development of intrinsic objectives, we retrospectively compute potential objectives on pre-collected datasets of agent behavior, rather than optimizing them online, and compare them by analyzing their correlations. We study input entropy, information gain, and empowerment across seven agents, three Atari games, and the 3D game Minecraft. We find that all three intrinsic objectives correlate more strongly with a human behavior similarity metric than with task reward. Moreover, input entropy and information gain correlate more strongly with human similarity than task reward does, suggesting the use of intrinsic objectives for designing agents that behave similarly to human players.
The Gamer's Private Network (GPN) is a client/server technology created by WTFast for making the network performance of online games faster and more reliable. GPN s use middle-mile servers and proprietary algorithms to better connect online video-game players to their game's servers across a wide-area network. Online games are a massive entertainment market and network latency is a key aspect of a player's competitive edge. This market means many different approaches to network architecture are implemented by different competing companies and that those architectures are constantly evolving. Ensuring the optimal connection between a client of WTFast and the online game they wish to play is thus an incredibly difficult problem to automate. Using machine learning, we analyzed historical network data from GPN connections to explore the feasibility of network latency prediction which is a key part of optimization. Our next step will be to collect live data (including client/server load, packet and port information and specific game state information) from GPN Minecraft servers and bots. We will use this information in a Reinforcement Learning model along with predictions about latency to alter the clients' and servers' configurations for optimal network performance. These investigations and experiments will improve the quality of service and reliability of GPN systems.
This paper introduces EvoCraft, a framework for Minecraft designed to study open-ended algorithms. We introduce an API that provides an open-source Python interface for communicating with Minecraft to place and track blocks. In contrast to previous work in Minecraft that focused on learning to play the game, the grand challenge we pose here is to automatically search for increasingly complex artifacts in an open-ended fashion. Compared to other environments used to study open-endedness, Minecraft allows the construction of almost any kind of structure, including actuated machines with circuits and mechanical components. We present initial baseline results in evolving simple Minecraft creations through both interactive and automated evolution. While evolution succeeds when tasked to grow a structure towards a specific target, it is unable to find a solution when rewarded for creating a simple machine that moves. Thus, EvoCraft offers a challenging new environment for automated search methods (such as evolution) to find complex artifacts that we hope will spur the development of more open-ended algorithms. A Python implementation of the EvoCraft framework is available at: https://github.com/real-itu/Evocraft-py.
In a search and rescue scenario, rescuers may have different knowledge of the environment and strategies for exploration. Understanding what is inside a rescuer's mind will enable an observer agent to proactively assist them with critical information that can help them perform their task efficiently. To this end, we propose to build models of the rescuers based on their trajectory observations to predict their strategies. In our efforts to model the rescuer's mind, we begin with a simple simulated search and rescue task in Minecraft with human participants. We formulate neural sequence models to predict the triage strategy and the next location of the rescuer. As the neural networks are data-driven, we design a diverse set of artificial "faux human" agents for training, to test them with limited human rescuer trajectory data. To evaluate the agents, we compare it to an evidence accumulation method that explicitly incorporates all available background knowledge and provides an intended upper bound for the expected performance. Further, we perform experiments where the observer/predictor is human. We show results in terms of prediction accuracy of our computational approaches as compared with that of human observers.
While there are many machine learning methods to classify and cluster sequences, they fail to explain what are the differences in groups of sequences that make them distinguishable. Although in some cases having a black box model is sufficient, there is a need for increased explainability in research areas focused on human behaviors. For example, psychologists are less interested in having a model that predicts human behavior with high accuracy and more concerned with identifying differences between actions that lead to divergent human behavior. This paper presents techniques for understanding differences between classes of discrete sequences. Approaches introduced in this paper can be utilized to interpret black box machine learning models on sequences. The first approach compares k-gram representations of sequences using the silhouette score. The second method characterizes differences by analyzing the distance matrix of subsequences. As a case study, we trained black box supervised learning methods to classify sequences of GitHub teams and then utilized our sequence analysis techniques to measure and characterize differences between event sequences of teams with bots and teams without bots. In our second case study, we classified Minecraft event sequences to infer their high-level actions and analyzed differences between low-level event sequences of actions.
A general-purpose planning agent requires an open-scope world model: one rich enough to tackle any of the wide range of tasks it may be asked to solve over its operational lifetime. This stands in contrast with typical planning approaches, where the scope of a model is limited to a specific family of tasks that share significant structure. Unfortunately, planning to solve any specific task using an open-scope model is computationally intractable - even for state-of-the-art methods - due to the many states and actions that are necessarily present in the model but irrelevant to that problem. We propose task scoping: a method that exploits knowledge of the initial state, goal conditions, and transition system to automatically and efficiently remove provably irrelevant variables and actions from a planning problem. Our approach leverages causal link analysis and backwards reachability over state variables (rather than states) along with operator merging (when effects on relevant variables are identical). Using task scoping as a pre-planning step can shrink the search space by orders of magnitude and dramatically decrease planning time. We empirically demonstrate that these improvements occur across a variety of open-scope domains, including Minecraft, where our approach leads to a 75x reduction in search time with a state-of-the-art numeric planner, even after including the time required for task scoping itself.
When generating technical instructions, it is often convenient to describe complex objects in the world at different levels of abstraction. A novice user might need an object explained piece by piece, while for an expert, talking about the complex object (e.g. a wall or railing) directly may be more succinct and efficient. We show how to generate building instructions at different levels of abstraction in Minecraft. We introduce the use of hierarchical planning to this end, a method from AI planning which can capture the structure of complex objects neatly. A crowdsourcing evaluation shows that the choice of abstraction level matters to users, and that an abstraction strategy which balances low-level and high-level object descriptions compares favorably to ones which don't.