How efficient is estimation with missing data?

Karadogan, Seliz G.; Marchegiani, Letizia; Hansen, Lars Kai; Larsen, Jan

doi:10.1109/icassp.2011.5946932

Cited by 10 publications

(13 citation statements)

References 10 publications

(10 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The regularization is based on inflating the diagonal elements similar to the approach presented in [11]. Further details can be found in [9]. We draw random samples from the distribution, µ 0 and Σ0 and apply the KKZ method [12].…”

Section: Modeling Frameworkmentioning

confidence: 99%

“…The model-based methods like, e.g., maximum Likelihood approaches [8], we do not modify or ignore part of the available data, but operate directly on the incomplete set. In [9] we evaluated a number of missing data techniques on classification problems, computing the misclassification rate (MR) for different missing data percentages (MDP) and proposing a new way to quantify their efficiency. In particular, we analyzed the behavior of mean imputation method, in which the missing variable is replaced with the mean value of the same variable, the pairwise deletion method, which consists of removing only the missing elements, and the maximum likelihood method in [8] (Complete EM).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Top-down attentionwith features missing at random

Karadogan

Marchegiani

Larsen

et al. 2011

2011 IEEE International Workshop on Machine Learning for Signal Processing

Self Cite

View full text Add to dashboard Cite

In this paper we present a top-down attention model designed for an environment in which features are missing completely at random. Following we model top-down attention as a sequential decision making process driven by a task -modeled as a classification problem -in an environment with random subsets of features missing, but where we have the possibility to gather additional features among the ones that are missing. Thus, the top-down attention problem is reduced to finding the answer to the question what to measure next? Attention is based on the top-down saliency of the missing features given as the estimated difference in classification confusion (entropy) with and without the given feature. The difference in confusion is computed conditioned on the available set of features. In this work, we make our attention model more realistic by also allowing the initial training phase to take place with incomplete data. Thus, we expand the model to include a missing data technique in the learning process. The top-down attention mechanism is implemented in a Gaussian Discrete mixture model setting where marginals and conditionals are relatively easy to compute. To illustrate the viability of expanded model, we train the mixture model with two different datasets, a synthetic data set and the well-known Yeast dataset of the UCI database. We evaluate the new algorithm in environments characterized by different amounts of incompleteness and compare the performance with a system that decides next feature to be measured at random. The proposed top-down mechanism clearly outperforms random choice of the next feature.

show abstract

Section: Modeling Frameworkmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Top-down attentionwith features missing at random

Karadogan

Marchegiani

Larsen

et al. 2011

2011 IEEE International Workshop on Machine Learning for Signal Processing

Self Cite

View full text Add to dashboard Cite

show abstract

“…In particular, we consider a scenario where the wearable device streams raw acceleration data using the undirected connectionless BLE advertisements (similarly to infrastructure presented in [10]). Although data reliability can be addressed at the receiver [26], this communication approach does not provide delivery guarantees and, thus, can be only applied to applications that can tolerate data loss or make use of specific missing data techniques [16]. We also assume the following.…”

Section: Power Profile and Battery Lifetime Estimationsmentioning

confidence: 99%

Designing Wearable Sensing Platforms for Healthcare in a Residential Environment

Fafoutis¹,

Vafeas²,

Janko

et al. 2017

EAI Endorsed Trans Perv Health Tech

View full text Add to dashboard Cite

Wearable technologies are valuable tools that can encourage people to monitor their own well-being and facilitate timely health interventions. In this paper, we present SPW-2; a low-profile versatile wearable sensor that employs two ultra low power accelerometers and an optional gyroscope. Designed for minimum maintenance and a long-term operation outside the laboratory, SPW-2 is able to offer a battery lifetime of multiple months. Measurements on its wireless performance in a real residential environment with thick brick walls, demonstrate that SPW-2 can fully cover a room and -in most cases -the adjacent room, as well.

show abstract

“…In particular, we consider a scenario where SPW-1 streams raw accelerometer data using the undirected connectionless BLE advertisements (similarly to [10]). Although data reliability can be addressed at the receiver [22], this communication approach does not provide delivery guarantees and, thus, can be only applied to applications that can tolerate data loss or make use of specific missing data techniques [13]. We also assume the following.…”

Section: Energy Consumption and Battery Lifetime Estimationsmentioning

confidence: 99%

SPW-1: A Low-Maintenance Wearable Activity Tracker for Residential Monitoring and Healthcare Applications

Fafoutis

Janko

Mellios

et al. 2016

Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering

View full text Add to dashboard Cite

Abstract. In this paper, we present SPW-1; a low-profile versatile wearable activity tracker that employs two ultra-low-power accelerometers and relies on Bluetooth Low Energy (BLE) for wireless communication. Aiming for a low maintenance system, SPW-1 is able to offer a battery lifetime of multiple months. Measurements on its wireless performance in a real residential environment with thick brick walls, demonstrate that SPW-1 can fully cover a room and -in most cases -the adjacent room, as well. SPW-1 is a research platform that is aimed to be used both as a data collecting tool for health-oriented studies outside the laboratory, but also for research on wearable technologies and body-centric communications. As a result, SPW-1 incorporates versatile features, such as external sensor support, various powering options, and accelerometer configuration options that can support a wide range applications from kinematics to long-term activity recognition.

show abstract

How efficient is estimation with missing data?

Cited by 10 publications

References 10 publications

Top-down attentionwith features missing at random

Top-down attentionwith features missing at random

Designing Wearable Sensing Platforms for Healthcare in a Residential Environment

SPW-1: A Low-Maintenance Wearable Activity Tracker for Residential Monitoring and Healthcare Applications

Contact Info

Product

Resources

About