Martin Azizyan

SurroundSense: mobile phone localization via ambience fingerprinting

A growing number of mobile computing applications are centered around the users location. The notion of location is broad, ranging from physical coordinates (latitude/longitude) to logical labels (like Starbucks, McDonalds). While extensive research has been performed in physical localization, there have been few attempts in recognizing logical locations. This paper argues that the increasing number of sensors on mobile phones presents new opportunities for logical localization. We postulate that ambient sound, light, and color in a place convey a photo-acoustic signature that can be sensed by the phones camera and microphone. In-built accelerometers in some phones may also be useful in inferring broad classes of user-motion, often dictated by the nature of the place. By combining these optical, acoustic, and motion attributes, it may be feasible to construct an identifiable fingerprint for logical localization. Hence, users in adjacent stores can be separated logically, even when their physical positions are extremely close. We propose SurroundSense, a mobile phone based system that explores logical localization via ambience fingerprinting. Evaluation results from 51 different stores show that SurroundSense can achieve an average accuracy of 87% when all sensing modalities are employed. We believe this is an encouraging result, opening new possibilities in indoor localization.

RNA design rules from a massive open laboratory

Significance Self-assembling RNA molecules play critical roles throughout biology and bioengineering. To accelerate progress in RNA design, we present EteRNA, the first internet-scale citizen science “game” scored by high-throughput experiments. A community of 37,000 nonexperts leveraged continuous remote laboratory feedback to learn new design rules that substantially improve the experimental accuracy of RNA structure designs. These rules, distilled by machine learning into a new automated algorithm EteRNABot, also significantly outperform prior algorithms in a gauntlet of independent tests. These results show that an online community can carry out large-scale experiments, hypothesis generation, and algorithm design to create practical advances in empirical science. Self-assembling RNA molecules present compelling substrates for the rational interrogation and control of living systems. However, imperfect in silico models—even at the secondary structure level—hinder the design of new RNAs that function properly when synthesized. Here, we present a unique and potentially general approach to such empirical problems: the Massive Open Laboratory. The EteRNA project connects 37,000 enthusiasts to RNA design puzzles through an online interface. Uniquely, EteRNA participants not only manipulate simulated molecules but also control a remote experimental pipeline for high-throughput RNA synthesis and structure mapping. We show herein that the EteRNA community leveraged dozens of cycles of continuous wet laboratory feedback to learn strategies for solving in vitro RNA design problems on which automated methods fail. The top strategies—including several previously unrecognized negative design rules—were distilled by machine learning into an algorithm, EteRNABot. Over a rigorous 1-y testing phase, both the EteRNA community and EteRNABot significantly outperformed prior algorithms in a dozen RNA secondary structure design tests, including the creation of dendrimer-like structures and scaffolds for small molecule sensors. These results show that an online community can carry out large-scale experiments, hypothesis generation, and algorithm design to create practical advances in empirical science.

SurroundSense: mobile phone localization using ambient sound and light

Proliferating mobile phones provide a foundation for revolutionary innovations in peoplecentric computing. Numerous applications are on the rise, many of which exploit the phones location as the primary indicator of context. We argue that existing physical localization schemes based on GPS/WiFi/GSM have limitations which make them impractical for use in such applications. Instead, in this poster we describe a means of localization where phones sense their surroundings, and use this ambient information to classify their location. Put differently, we postulate that different surroundings have photo-acoustic fingerprints, that can be sensed and used for localization. We demonstrate the feasibility using Tmote Invent motes that have light and sound sensors. Our ongoing work is extending SurroundSense to the mobile phone platform, and exploiting additional sensors (such as accelerometers and compasses) towards even better localization.

Subspace learning from extremely compressed measurements

We consider learning the principal subspace of a large set of vectors from an extremely small number of compressive measurements of each vector. Our theoretical results show that even a constant number of measurements per column suffices to approximate the principal subspace to arbitrary precision, provided that the number of vectors is large. This result is achieved by a simple algorithm that computes the eigenvectors of an estimate of the covariance matrix. The main insight is to exploit an averaging effect that arises from applying a different random projection to each vector. We provide a number of simulations confirming our theoretical results.

Density-sensitive semisupervised inference

Semisupervised methods are techniques for using labeled data

Subspace detection of high-dimensional vectors using compressive sampling

(X_1,Y_1),\ldots,(X_n,Y_n)

Did you see Bob?: human localization using mobile phones

together with unlabeled data

Minimax Theory for High-dimensional Gaussian Mixtures with Sparse Mean Separation

X_{n+1},\ldots,X_N

Feature Selection For High-Dimensional Clustering

to make predictions. These methods invoke some assumptions that link the marginal distribution

Extreme Compressive Sampling for Covariance Estimation

P_X