Dominant Particle Agent (DPA) google
We describe a new approach for mitigating risk in the Reinforcement Learning paradigm. Instead of reasoning about expected utility, we use second-order stochastic dominance (SSD) to directly compare the inherent risk of random returns induced by different actions. We frame the RL optimization within the space of probability measures to accommodate the SSD relation, treating Bellman’s equation as a potential energy functional. This brings us to Wasserstein gradient flows, for which the optimality and convergence are well understood. We propose a discrete-measure approximation algorithm called the Dominant Particle Agent (DPA), and we demonstrate how safety and performance are better balanced with DPA than with existing baselines. …

Distance to Kernel and Embedding (D2KE) google
For many machine learning problem settings, particularly with structured inputs such as sequences or sets of objects, a distance measure between inputs can be specified more naturally than a feature representation. However, most standard machine models are designed for inputs with a vector feature representation. In this work, we consider the estimation of a function $f:\mathcal{X} \rightarrow \R$ based solely on a dissimilarity measure $d:\mathcal{X}\times\mathcal{X} \rightarrow \R$ between inputs. In particular, we propose a general framework to derive a family of \emph{positive definite kernels} from a given dissimilarity measure, which subsumes the widely-used \emph{representative-set method} as a special case, and relates to the well-known \emph{distance substitution kernel} in a limiting case. We show that functions in the corresponding Reproducing Kernel Hilbert Space (RKHS) are Lipschitz-continuous w.r.t. the given distance metric. We provide a tractable algorithm to estimate a function from this RKHS, and show that it enjoys better generalizability than Nearest-Neighbor estimates. Our approach draws from the literature of Random Features, but instead of deriving feature maps from an existing kernel, we construct novel kernels from a random feature map, that we specify given the distance measure. We conduct classification experiments with such disparate domains as strings, time series, and sets of vectors, where our proposed framework compares favorably to existing distance-based learning methods such as $k$-nearest-neighbors, distance-substitution kernels, pseudo-Euclidean embedding, and the representative-set method. …

Approximate Query Processing (AQP) google
While Big Data opens the possibility of gaining unprecedented insights, it comes at the price of increased need for computational resources (or risk of higher latency) for answering queries over voluminous data. The ability to provide approximate answers to queries at a fraction of the cost of executing the query in the traditional way, has the disruptive potential of allowing us to explore large datasets efficiently. Specifically, such techniques could prove effective in helping data scientists identify the subset of data that needs further drill-down, discovering trends, and enabling fast visualization. In this article, we will focus on approximate query processing schemes that are based on sampling techniques. An approximate query processing (AQP) scheme can be characterized by the generality of the query language it supports, its error model and accuracy guarantee, the amount of work saved at runtime, and the amount of additional resources it requires in precomputation. These dimensions of an AQP scheme are not independent and much of the past work makes specific choices along the above four dimensions. Specifically, it seems impossible to have an AQP system that supports the richness of SQL with significant saving of work while providing an accuracy guarantee that is acceptable to a broad set of application workloads. Put another way, you cannot have it all.
MISS: Finding Optimal Sample Sizes for Approximate Analytics
Approximate Query Processing using Deep Generative Models

Aggregate Valuation of Antecedents (AVA) google
Current approaches for explaining machine learning models fall into two distinct classes: antecedent event influence and value attribution. The former leverages training instances to describe how much influence a training point exerts on a test point, while the latter attempts to attribute value to the features most pertinent to a given prediction. In this work, we discuss an algorithm, AVA: Aggregate Valuation of Antecedents, that fuses these two explanation classes to form a new approach to feature attribution that not only retrieves local explanations but also captures global patterns learned by a model. Our experimentation convincingly favors weighting and aggregating feature attributions via AVA. …