Procedural Content Generation (PCG) using machine learning is a fast growing area of research. Action Role Playing Game (ARPG) levels represent an interesting challenge for PCG due to their multi-tiered structure and nonlinearity. Previous work has used Bayes Nets (BN) to learn properties of the topological structure of levels from The Legend of Zelda. In this paper we describe a method for sampling these learned distributions to generate valid, playable level topologies. We carry this deeper and learn a sampleable representation of the individual rooms using Principal Component Analysis. We combine the two techniques and present a multi-scale machine learned technique for procedurally generating ARPG levels from a corpus of levels from The Legend of Zelda.

Behavioral game analytics has predominantly been confined to work on single games, which means that the cross-game applicability of current knowledge remains largely unknown. Here four experiments are presented focusing on the relationship between game ownership, time invested in playing games, and the players themselves, across more than 3000 games distributed by the Steam platform and over 6 million players, covering a total playtime of over 5 billion hours. Experiments are targeted at uncovering high-level patterns in the behavior of players focusing on playtime, using frequent itemset mining on game ownership, cluster analysis to develop playtime-dependent player profiles, correlation between user game rankings and, review scores, playtime and game ownership, as well as cluster analysis on Steam games. Within the context of playtime, the analyses presented provide unique insights into the behavior of game players as they occur across games, for example in how players distribute their time across games.

The growing need to analyze large collections of documents has led to great developments in topic modeling. Since documents are frequently associated with other related variables, such as labels or ratings, much interest has been placed on supervised topic models. However, the nature of most annotation tasks, prone to ambiguity and noise, often with high volumes of documents, deem learning under a single-annotator assumption unrealistic or unpractical for most real-world applications. In this paper, we propose a supervised topic model that accounts for the heterogeneity and biases among different annotators that are encountered in practice when learning from crowds. We develop an efficient stochastic variational inference algorithm that is able to scale to very large datasets, and we empirically demonstrate the advantages of the proposed model over state of the art approaches.

Mobile digital games are dominantly released under the freemium business model, but only a small fraction of the players makes any purchases. The ability to predict who will make a purchase enables optimization of marketing efforts, and tailoring customer relationship management to the specific user's profile. Here this challenge is addressed via two models for predicting purchasing players, using a 100,000 player dataset: 1) A classification model focused on predicting whether a purchase will occur or not. 2) a regression model focused on predicting the number of purchases a user will make. Both models are presented within a decision and regression tree framework for building rules that are actionable by companies. To the best of our knowledge, this is the first study investigating purchase decisions in freemium mobile products from a user behavior perspective and adopting behavior-driven learning approaches to this problem.

Counting objects is a fundamental image processisng primitive, and has many scientific, health, surveillance, security, and military applications. Existing supervised computer vision techniques typically require large quantities of labeled training data, and even with that, fail to return accurate results in all but the most stylized settings. Using vanilla crowdsourcing, on the other hand, can lead to significant errors, especially on images with many objects. In this paper, we present our JellyBean suite of algorithms, that combines the best of crowds and computer vision to count objects in images, and uses judicious decomposition of images to greatly improve accuracy at low cost. Our algorithms have several desirable properties: (i) they are theoretically optimal or near-optimal , in that they ask as few questions as possible to humans (under certain intuitively reasonable assumptions that we justify in our paper experimentally); (ii) they operate under stand-alone or hybrid modes, in that they can either work independent of computer vision algorithms, or work in concert with them, depending on whether the computer vision techniques are available or useful for the given setting; (iii) they perform very well in practice, returning accurate counts on images that no individual worker or computer vision algorithm can count correctly, while not incurring a high cost.

While recent work has shown that a worker’s performance can be more accurately modeled by temporal correlation in task performance, a fundamental challenge remains in the need for expert gold labels to evaluate a worker’s performance. To solve this problem, we explore two methods of utilizing limited gold labels, initial training and periodic updating. Furthermore, we present a novel way of learning a prediction model in the absence of gold labels with uncertaintyaware learning and soft-label updating. Our experiment with a real crowdsourcing dataset demonstrates that periodic updating tends to show better performance than initial training when the number of gold labels are very limited (< 25).

We report on a series of behavioral experiments in human computation on three different tasks over networks: graph coloring, community detection (or graph clustering), and competitive contagion. While these tasks share similar action spaces and interfaces, they capture a diversity of computational challenges: graph coloring is a search problem, clustering is an optimization problem, and competitive contagion is a game-theoretic problem. In contrast with most of the prior literature on human-subject experiments in networks, in which collectives of subjects are embedded "in" the network, and have only local information and interactions, here individual subjects have a global (or "over") view and must solve "whole network" problems alone. Our primary findings are that subject performance is impressive across all three problem types; that subjects find diverse and novel strategies for solving each task; and that collective performance can often be strongly correlated with known algorithms.

We consider a finite-pool data categorization scenario which requires exhaustively classifying a given set of examples with a limited budget. We adopt a hybrid human-machine approach which blends automatic machine learning with human labeling across a tiered workforce composed of domain experts and crowd workers. To effectively achieve high-accuracy labels over the instances in the pool at minimal cost, we develop a novel approach based on decision-theoretic active learning. On the important task of biomedical citation screening for systematic reviews, results on real data show that our method achieves consistent improvements over baseline strategies. To foster further research by others, we have made our data available online.

Models for aggregating contributions by crowd workers have been shown to be challenged by the rise of task-specific biases or errors. Task-dependent errors in assessment may shift the majority opinion of even large numbers of workers to an incorrect answer. We introduce and evaluate probabilistic models that can detect and correct task-dependent bias automatically. First, we show how to build and use probabilistic graphical models for jointly modeling task features, workers' biases, worker contributions and ground truth answers of tasks so that task-dependent bias can be corrected. Second, we show how the approach can perform a type of transfer learning among workers to address the issue of annotation sparsity. We evaluate the models with varying complexity on a large data set collected from a citizen science project and show that the models are effective at correcting the task-dependent worker bias. Finally, we investigate the use of active learning to guide the acquisition of expert assessments to enable automatic detection and correction of worker bias.

Stochastic gradient algorithms have been the main focus of large-scale learning problems and they led to important successes in machine learning. The convergence of SGD depends on the careful choice of learning rate and the amount of the noise in stochastic estimates of the gradients. In this paper, we propose a new adaptive learning rate algorithm, which utilizes curvature information for automatically tuning the learning rates. The information about the element-wise curvature of the loss function is estimated from the local statistics of the stochastic first order gradients. We further propose a new variance reduction technique to speed up the convergence. In our preliminary experiments with deep neural networks, we obtained better performance compared to the popular stochastic gradient algorithms.