Source author record

Wendi Heinzelman

Wendi Heinzelman appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

Human-Computer Interaction Networking and Internet Architecture

Catalog footprint

What is connected

2works

2topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2016arXiv

Emotion Classification: How Does an Automated System Compare to Naive Human Coders?

The fact that emotions play a vital role in social interactions, along with the demand for novel human-computer interaction applications, have led to the development of a number of automatic emotion classification systems. However, it is still debatable whether the performance of such systems can compare with human coders. To address this issue, in this study, we present a comprehensive comparison in a speech-based emotion classification task between 138 Amazon Mechanical Turk workers (Turkers) and a state-of-the-art automatic computer system. The comparison includes classifying speech utterances into six emotions (happy, neutral, sad, anger, disgust and fear), into three arousal classes (active, passive, and neutral), and into three valence classes (positive, negative, and neutral). The results show that the computer system outperforms the naive Turkers in almost all cases. Furthermore, the computer system can increase the classification accuracy by rejecting to classify utterances for which it is not confident, while the Turkers do not show a significantly higher classification accuracy on their confident utterances versus unconfident ones.

preprint2014arXiv

Extending the Energy Framework for Network Simulator 3 (ns-3)

The problem of designing and simulating optimal transmission protocols for energy harvesting wireless networks has recently received considerable attention, thus requiring for an accurate modeling of the energy harvesting process and a consequent redesign of the simulation framework to include it. While the current ns-3 energy framework allows the definition of new energy sources that incorporate the contribution of an energy harvester, the integration of an energy harvester component into an existing energy source is not straightforward using the existing energy framework. In this poster, we propose an extension of the energy framework currently released with ns-3 in order to explicitly introduce the concept of an energy harvester. Starting from the definition of the general interface, we then provide the implementation of two simple models for the energy harvester. In addition, we extend the set of implementations of the current energy framework to include a model for a supercapacitor energy source and a device energy model for the energy consumption of a sensor. Finally, we introduce the concept of an energy predictor, that gathers information from the energy source and harvester and use this information to predict the amount of energy that will be available in the future, and we provide an example implementation. As a result of these efforts, we believe that our contributions to the ns-3 energy framework will provide a useful tool to enhance the quality of simulations of energy-aware wireless networks.