Christian R. Shelton

Fast time series classification using numerosity reduction

Many algorithms have been proposed for the problem of time series classification. However, it is clear that one-nearest-neighbor with Dynamic Time Warping (DTW) distance is exceptionally difficult to beat. This approach has one weakness, however; it is computationally too demanding for many realtime applications. One way to mitigate this problem is to speed up the DTW calculations. Nonetheless, there is a limit to how much this can help. In this work, we propose an additional technique, numerosity reduction, to speed up one-nearest-neighbor DTW. While the idea of numerosity reduction for nearest-neighbor classifiers has a long history, we show here that we can leverage off an original observation about the relationship between dataset size and DTW constraints to produce an extremely compact dataset with little or no loss in accuracy. We test our ideas with a comprehensive set of experiments, and show that it can efficiently produce extremely fast accurate classifiers.

Momentum control for balance

We demonstrate a real-time simulation system capable of automatically balancing a standing character, while at the same time tracking a reference motion and responding to external perturbations. The system is general to non-human morphologies and results in natural balancing motions employing the entire body (for example, wind-milling). Our novel balance routine seeks to control the linear and angular momenta of the character. We demonstrate how momentum is related to the center of mass and center of pressure of the character and derive control rules to change these centers for balance. The desired momentum changes are reconciled with the objective of tracking the reference motion through an optimization routine which produces target joint accelerations. A hybrid inverse/forward dynamics algorithm determines joint torques based on these joint accelerations and the ground reaction forces. Finally, the joint torques are applied to the free-standing character simulation. We demonstrate results for following both motion capture and keyframe data as well as both human and non-human morphologies in presence of a variety of conditions and disturbances.

Annotating Genes of Known and Unknown Function by Large-Scale Coexpression Analysis

About 40% of the proteins encoded in eukaryotic genomes are proteins of unknown function (PUFs). Their functional characterization remains one of the main challenges in modern biology. In this study we identified the PUF encoding genes from Arabidopsis (Arabidopsis thaliana) using a combination of sequence similarity, domain-based, and empirical approaches. Large-scale gene expression analyses of 1,310 publicly available Affymetrix chips were performed to associate the identified PUF genes with regulatory networks and biological processes of known function. To generate quality results, the study was restricted to expression sets with replicated samples. First, genome-wide clustering and gene function enrichment analysis of clusters allowed us to associate 1,541 PUF genes with tightly coexpressed genes for proteins of known function (PKFs). Over 70% of them could be assigned to more specific biological process annotations than the ones available in the current Gene Ontology release. The most highly overrepresented functional categories in the obtained clusters were ribosome assembly, photosynthesis, and cell wall pathways. Interestingly, the majority of the PUF genes appeared to be controlled by the same regulatory networks as most PKF genes, because clusters enriched in PUF genes were extremely rare. Second, large-scale analysis of differentially expressed genes was applied to identify a comprehensive set of abiotic stress-response genes. This analysis resulted in the identification of 269 PKF and 104 PUF genes that responded to a wide variety of abiotic stresses, whereas 608 PKF and 206 PUF genes responded predominantly to specific stress treatments. The provided coexpression and differentially expressed gene data represent an important resource for guiding future functional characterization experiments of PUF and PKF genes. Finally, the public Plant Gene Expression Database (http://bioweb.ucr.edu/PED) was developed as part of this project to provide efficient access and mining tools for the vast gene expression data of this study.

Improving multi-target tracking via social grouping

We address the problem of multi-person data-association-based tracking (DAT) in semi-crowded environments from a single camera. Existing tracklet-association-based methods using purely visual cues (like appearance and motion information) show impressive results but rely on heavy training, a number of tuned parameters, and sophisticated detectors to cope with visual ambiguities within the video and low-level processing errors. In this work, we consider clustering dynamics to mitigate such ambiguities. This leads to a general optimization framework that adds social grouping behavior (SGB) to any basic affinity model. We formulate this as a nonlinear global optimization problem to maximize the consistency of visual and grouping cues for trajectories in both tracklet-tracklet linking space and tracklet-grouping assignment space. We formulate the Lagrange dual and solve it using a two-stage iterative algorithm, employing the Hungarian algorithm and K-means clustering. We build SGB upon a simple affinity model and show very promising performance on two publicly available real-world datasets with different tracklet extraction methods.

A social reinforcement learning agent

We report on our reinforcement learning work on Cobot, a software agent that resides in the well-known online chat community LambdaMOO. Our initial work on Cobot~\cite{cobotaaai} provided him with the ability to collect {\em social statistics\/} and report them to users in a reactive manner. Here we describe our application of reinforcement learning to allow Cobot to proactively take actions in this complex social environment, and adapt his behavior from multiple sources of human reward. After 5 months of training, Cobot received 3171 reward and punishment events from 254 different Lambda\-MOO users, and learned nontrivial preferences for a number of users. Cobot modifies his behavior based on his current state in an attempt to maximize reward. Here we describe LambdaMOO and the state and action spaces of Cobot, and report the statistical results of the learning experiment.

Morphable Surface Models

We describe a novel automatic technique for finding a dense correspondence between a pair of n-dimensional surfaces with arbitrary topologies. This method employs a different formulation than previous correspondence algorithms (such as optical flow) and includes images as a special case. We use this correspondence algorithm to build Morphable Surface Models (an extension of Morphable Models) from examples. We present a method for matching the model to new surfaces and demonstrate their use for analysis, synthesis, and clustering.

A continuation method for Nash equilibria in structured games

Structured game representations have recently attracted interest as models for multi-agent artificial intelligence scenarios, with rational behavior most commonly characterized by Nash equilibria. This paper presents efficient, exact algorithms for computing Nash equilibria in structured game representations, including both graphical games and multi-agent influence diagrams (MAIDs). The algorithms are derived from a continuation method for normal-form and extensive-form games due to Govindan and Wilson; they follow a trajectory through a space of perturbed games and their equilibria, exploiting game structure through fast computation of the Jacobian of the payoff function. They are theoretically guaranteed to find at least one equilibrium of the game, and may find more. Our approach provides the first efficient algorithm for computing exact equilibria in graphical games with arbitrary topology, and the first algorithm to exploit fine-grained structural properties of MAIDs. Experimental results are presented demonstrating the effectiveness of the algorithms and comparing them to predecessors. The running time of the graphical game algorithm is similar to, and often better than, the running time of previous approximate algorithms. The algorithm for MAIDs can effectively solve games that are much larger than those solvable by previous methods.

Automated, highly-accurate, bug assignment using machine learning and tossing graphs

Empirical studies indicate that automating the bug assignment process has the potential to significantly reduce software evolution effort and costs. Prior work has used machine learning techniques to automate bug assignment but has employed a narrow band of tools which can be ineffective in large, long-lived software projects. To redress this situation, in this paper we employ a comprehensive set of machine learning tools and a probabilistic graph-based model (bug tossing graphs) that lead to highly-accurate predictions, and lay the foundation for the next generation of machine learning-based bug assignment. Our work is the first to examine the impact of multiple machine learning dimensions (classifiers, attributes, and training history) along with bug tossing graphs on prediction accuracy in bug assignment. We validate our approach on Mozilla and Eclipse, covering 856,259 bug reports and 21 cumulative years of development. We demonstrate that our techniques can achieve up to 86.09% prediction accuracy in bug assignment and significantly reduce tossing path lengths. We show that for our data sets the Naive Bayes classifier coupled with product-component features, tossing graphs and incremental learning performs best. Next, we perform an ablative analysis by unilaterally varying classifiers, features, and learning model to show their relative importance of on bug assignment accuracy. Finally, we propose optimization techniques that achieve high prediction accuracy while reducing training and prediction time.

Intrusion detection using continuous time Bayesian networks

Intrusion detection systems (IDSs) fall into two high-level categories: network-based systems (NIDS) that monitor network behaviors, and host-based systems (HIDS) that monitor system calls. In this work, we present a general technique for both systems. We use anomaly detection, which identifies patterns not conforming to a historic norm. In both types of systems, the rates of change vary dramatically over time (due to burstiness) and over components (due to service difference). To efficiently model such systems, we use continuous time Bayesian networks (CTBNs) and avoid specifying a fixed update interval common to discrete-time models. We build generative models from the normal training data, and abnormal behaviors are flagged based on their likelihood under this norm. For NIDS, we construct a hierarchical CTBN model for the network packet traces and use Rao-Blackwellized particle filtering to learn the parameters. We illustrate the power of our method through experiments on detecting real worms and identifying hosts on two publicly available network traces, the MAWI dataset and the LBNL dataset. For HIDS, we develop a novel learning method to deal with the finite resolution of system log file time stamps, without losing the benefits of our continuous time model. We demonstrate the method by detecting intrusions in the DARPA 1998 BSM dataset.

Cobot in LambdaMOO: An Adaptive Social Statistics Agent

We describe our development of Cobot, a novel software agent who lives in LambdaMOO, a popular virtual world frequented by hundreds of users. Cobot’s goal was to become an actual part of that community. Here, we present a detailed discussion of the functionality that made him one of the objects most frequently interacted with in LambdaMOO, human or artificial. Cobot’s fundamental power is that he has the ability to collect social statistics summarizing the quantity and quality of interpersonal interactions. Initially, Cobot acted as little more than a reporter of this information; however, as he collected more and more data, he was able to use these statistics as models that allowed him to modify his own behavior. In particular, cobot is able to use this data to “self-program,” learning the proper way to respond to the actions of individual users, by observing how others interact with one another. Further, Cobot uses reinforcement learning to proactively take action in this complex social environment, and adapts his behavior based on multiple sources of human reward. Cobot represents a unique experiment in building adaptive agents who must live in and navigate social spaces.