Identification of falls while performing normal activities of daily living (ADL) is important to ensure personal safety and well-being. However, falling is a short term activity that occurs infrequently. This poses a challenge to traditional classification algorithms, because there may be very little training data for falls (or none at all). This paper proposes an approach for the identification of falls using a wearable device in the absence of training data for falls but with plentiful data for normal ADL. We propose three `X-Factor' Hidden Markov Model (XHMMs) approaches. The XHMMs model unseen falls using "inflated" output covariances (observation models). To estimate the inflated covariances, we propose a novel cross validation method to remove "outliers" from the normal ADL that serve as proxies for the unseen falls and allow learning the XHMMs using only normal activities. We tested the proposed XHMM approaches on two activity recognition datasets and show high detection rates for falls in the absence of fall-specific training data. We show that the traditional method of choosing a threshold based on maximum of negative of log-likelihood to identify unseen falls is ill-posed for this problem. We also show that supervised classification methods perform poorly when very limited fall data are available during the training phase.

In this paper, we bring techniques from operations research to bear on the problem of choosing optimal actions in partially observable stochastic domains. We begin by introducing the theory of Markov decision processes (mdps) and partially observable MDPs (pomdps). We then outline a novel algorithm for solving pomdps off line and show how, in some cases, a finite-memory controller can be extracted from the solution to a POMDP. We conclude with a discussion of how our approach relates to previous work, the complexity of finding exact solutions to pomdps, and of some possibilities for finding approximate solutions.

We introduce a new dynamical system for sequentially observed multivariate count data. This model is based on the gamma--Poisson construction---a natural choice for count data---and relies on a novel Bayesian nonparametric prior that ties and shrinks the model parameters, thus avoiding overfitting. We present an efficient MCMC inference algorithm that advances recent work on augmentation schemes for inference in negative binomial models. Finally, we demonstrate the model's inductive bias using a variety of real-world data sets, showing that it exhibits superior predictive performance over other models and infers highly interpretable latent structure.

This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

The problem with using the human voice to control computers is well known and well documented: it doesn't always work. You can find yourself adopting the aggressive tone of a belligerent tourist in a foreign land while digital assistants employ a range of apologetic responses ("I'm sorry, I didn't quite get that", "I'm sorry, I didn't understand the question"). We throw our arms up and complain about their shortcomings. Plenty of us have tried them, plenty of us have dismissed them as a waste of time. We tend not to hear about them doing the job perfectly well, because few people write impassioned tweets or blog posts about things that work flawlessly.

The Teachable Agents Project combines research from computer science, psychology, and education to develop computer-based learning environments. These environments utilize animated pedagogical agents to facilitate science learning and the development of self-regulated learning skills. The use of animated agents allows us to extend the cognitive scaffolding provided by various computer tools and representations (e.g., searchable text, simulations, concept maps, etc.) by embedding them in productive and motivating social-constructive interactions (e.g., peer teaching, collaboration, and assessment). Current projects include Betty's Brain, a learning-by-teaching environment for science learning; CTSiM, an environment for understanding science through a computational thinking framework; SimSelf, a relatively new project that focuses on teaching students about self-regulation and metacognition in the context of science learning; and C3STEM, a community-situated, challenge-based, collaborative STEM learning environment. Our learning environments also include extensive logging of students' interactions with the system and agents.

A system being developed by Christopher Amato, a postdoc at MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL), can perform security-camera analysis to identify potential terrorists or illegal entry more accurately and in a fraction of the time it would take a human camera operator. "You can't have a person staring at every single screen, and even if you did the person might not know exactly what to look for," Amato says. "For example, a person is not going to be very good at searching through pages and pages of faces to try to match [an intruder] with a known criminal or terrorist." Existing computer vision systems designed to carry out this task automatically tend to be fairly slow, Amato says. "Sometimes it's important to come up with an alarm immediately, even if you are not yet positive exactly what it is happening," he says.

"Mahout" is a Hindi term for a person who rides an elephant. The elephant, in this case, is Hadoop -- and Mahout is one of the many projects that can sit on top of Hadoop, although you do not always need MapReduce to run it. Mahout puts powerful mathematical tools in the hands of the mere mortal developers who write the InterWebs. It's a package of implementations of the most popular and important machine-learning algorithms, with the majority of the implementations designed specifically to use Hadoop to enable scalable processing of huge data sets. Some algorithms are available only in a nonparallelizable "serial" form due to the nature of the algorithm, but all can take advantage of HDFS for convenient access to data in your Hadoop processing pipeline.

Here is an excellent ILP reference, somewhat older but publicly available. The most recent issue of SIGKDD Explorations is a special issue devoted to multi-relational data mining.

This page contains some information on research by David Poole and students on probabilistic reasoning and decision making. It is not intended to be an introduction to the vast literature on these topics, but only the incremental work done by me. For more different perspectives, see the pointers from the Uncertainty in AI (UAI) home page. Maybe someday I will write an online introduction. Probabilistic Horn abduction is a pragmatic combination of logic and probability.