Yexiang Xue

Same-decision probability: A confidence measure for threshold-based decisions

We consider in this paper the robustness of decisions based on probabilistic thresholds. To this effect, we propose the same-decision probability as a query that can be used as a confidence measure for threshold-based decisions. More specifically, the same-decision probability is the probability that we would have made the same threshold-based decision, had we known the state of some hidden variables pertaining to our decision. We study a number of properties about the same-decision probability. First, we analyze its computational complexity. We then derive a bound on its value, which we can compute using a variable elimination algorithm that we propose. Finally, we consider decisions based on noisy sensors in particular, showing through examples that the same-decision probability can be used to reason about threshold-based decisions in a more refined way.

Learning policies for battery usage optimization in electric vehicles

The high cost, limited capacity, and long recharge time of batteries pose a number of obstacles for the widespread adoption of electric vehicles. Multi-battery systems that combine a standard battery with supercapacitors are currently one of the most promising ways to increase battery lifespan and reduce operating costs. However, their performance crucially depends on how they are designed and operated.In this paper, we formalize the problem of optimizing real-time energy management of multi-battery systems as a stochastic planning problem, and we propose a novel solution based on a combination of optimization, machine learning and data-mining techniques. We evaluate the performance of our intelligent energy management system on various large datasets of commuter trips crowdsourced in the United States. We show that our policy significantly outperforms the leading algorithms that were previously proposed as part of an open algorithmic challenge. Further, we show how to extend our approach to an incremental learning setting, where the policy is capable of improving and adapting as new data is being collected over time.

Deep Multi-species Embedding

Understanding how species are distributed across landscapes over time is a fundamental question in biodiversity research. Unfortunately, most species distribution models only target a single species at a time, despite strong ecological evidence that species are not independently distributed. We propose Deep Multi-Species Embedding (DMSE), which jointly embeds vectors corresponding to multiple species as well as vectors representing environmental covariates into a common high-dimensional feature space via a deep neural network. Applied to bird observational data from the citizen science project \textit{eBird}, we demonstrate how the DMSE model discovers inter-species relationships to outperform single-species distribution models (random forests and SVMs) as well as competing multi-label models. Additionally, we demonstrate the benefit of using a deep neural network to extract features within the embedding and show how they improve the predictive performance of species distribution modelling. An important domain contribution of the DMSE model is the ability to discover and describe species interactions while simultaneously learning the shared habitat preferences among species. As an additional contribution, we provide a graphical embedding of hundreds of bird species in the Northeast US.

Behavior Identification in Two-Stage Games for Incentivizing Citizen Science Exploration

We consider two-stage games in which a leader seeks to direct the activities of independent agents by offering incentives. A good leader’s strategy requires an understanding of the agents’ utilities and the ability to predict agent behavior. Moreover, the optimization of outcomes requires an agent behavior model that can be efficiently incorporated into the leader’s model. Here we address the agent behavior modeling problem and show how it can be used to reduce bias in a challenging citizen science application. Adapting ideas from Discrete Choice Modeling in behavioral economics, we develop a probabilistic behavioral model that takes into account variable patterns of human behavior and suboptimal actions. By modeling deviations from baseline behavior we are able to accurately predict future behavior based on limited, sparse data. We provide a novel scheme to fold the agent model into a bi-level optimization as a single Mixed Integer Program, and scale up our approach by adding redundant constraints, based on novel insights of an easy-hard-easy phase transition phenomenon. We apply our methodology to a game called Avicaching, in collaboration with eBird, a well-established citizen science program that collects bird observations for conservation. Field results show that our behavioral model performs well and that the incentives are remarkably effective at steering citizen scientists’ efforts to reduce bias by exploring under-sampled areas. Moreover, the data collected from Avicaching improves the performance of species distribution models.

Boosting Efficiency for Computing the Pareto Frontier on Tree Structured Networks

Multi-objective optimization plays a key role in the study of real-world problems, as they often involve multiple criteria. In multi-objective optimization it is important to identify the so-called Pareto frontier, which characterizes the trade-offs between the objectives of different solutions. We show how a divide-and-conquer approach, combined with batched processing and pruning, significantly boosts the performance of an exact and approximation dynamic programming (DP) algorithm for computing the Pareto frontier on tree-structured networks, proposed in [18]. We also show how exploiting restarts and a new instance selection strategy boosts the performance and accuracy of a mixed integer programming (MIP) approach for approximating the Pareto frontier. We provide empirical results demonstrating that our DP and MIP approaches have complementary strengths and outperform previous algorithms in efficiency and accuracy. Our work is motivated by a problem in computational sustainability concerning the evaluation of trade-offs in ecosystem services due to the proliferation of hydropower dams throughout the Amazon basin. Our approaches are general and can be applied to computing the Pareto frontier of a variety of multi-objective problems on tree-structured networks.

Relaxation Methods for Constrained Matrix Factorization Problems: Solving the Phase Mapping Problem in Materials Discovery

Matrix factorization is a robust and widely adopted technique in data science, in which a given matrix is decomposed as the product of low rank matrices. We study a challenging constrained matrix factorization problem in materials discovery, the so-called phase mapping problem. We introduce a novel “lazy” Iterative Agile Factor Decomposition (IAFD) approach that relaxes and postpones non-convex constraint sets (the lazy constraints), iteratively enforcing them when violations are detected. IAFD interleaves multiplicative gradient-based updates with efficient modular algorithms that detect and repair constraint violations, while still ensuring fast run times. Experimental results show that IAFD is several orders of magnitude faster and its solutions are also in general considerably better than previous approaches. IAFD solves a key problem in materials discovery while also paving the way towards tackling constrained matrix factorization problems in general, with broader implications for data science.

Phase Mapper: Accelerating Materials Discovery with AI

From the stone age, to the bronze, iron age, and modern silicon age, the discovery and characterization of new materials has always been instrumental to humanitys progress and development. With the current pressing need to address sustainability challenges and find alternatives to fossil fuels, we look for solutions in the development of new materials that will allow for renewable energy. To discover materials with the required properties, materials scientists can perform high-throughput materials discovery, which includes rapid synthesis and characterization via X-ray diffraction (XRD) of thousands of materials. A central problem in materials discovery, the phase map identification problem, involves the determination of the crystal structure of materials from materials composition and structural characterization data. This analysis is traditionally performed mainly by hand, which can take days for a single material system. In this work we present Phase-Mapper, a solution platform that tightly integrates XRD experimentation, AI problem solving, and human intelligence for interpreting XRD patterns and inferring the crystal structures of the underlying materials. Phase-Mapper is compatible with any spectral demixing algorithm, including our novel solver, AgileFD, which is based on convolutive non-negative matrix factorization. AgileFD allows materials scientists to rapidly interpret XRD patterns, and incorporates constraints to capture prior knowledge about the physics of the materials as well as human feedback. With our system, materials scientists have been able to interpret previously unsolvable systems of XRD data at the Department of Energy’s Joint Center for Artificial Photosynthesis, including the Nb-Mn-V oxide system, which led to the discovery of new solar light absorbers and is provided as an illustrative example of AI-enabled high throughput materials discovery.

XOR-Sampling for Network Design with Correlated Stochastic Events

Many network optimization problems can be formulated as stochastic network design problems in which edges are present or absent stochastically. Furthermore, protective actions can guarantee that edges will remain present. We consider the problem of finding the optimal protection strategy under a budget limit in order to maximize some connectivity measurements of the network. Previous approaches rely on the assumption that edges are independent. In this paper, we consider a more realistic setting where multiple edges are not independent due to natural disasters or regional events that make the states of multiple edges stochastically correlated. We use Markov Random Fields to model the correlation and define a new stochastic network design framework. We provide a novel algorithm based on Sample Average Approximation (SAA) coupled with a Gibbs or XOR sampler. The experimental results on real road network data show that the policies produced by SAA with the XOR sampler have higher quality and lower variance compared to SAA with Gibbs sampler.