Carnegie Mellon University

About: Carnegie Mellon University is a education organization based out in Pittsburgh, Pennsylvania, United States. It is known for research contribution in the topics: Computer science & Robot. The organization has 36317 authors who have published 104359 publications receiving 5975734 citations. The organization is also known as: CMU & Carnegie Mellon.

Semi-Supervised Self-Training of Object Detection Models

Abstract: The construction of appearance-based object detection systems is time-consuming and difficult because a large number of training examples must be collected and manually labeled in order to capture variations in object appearance. Semi-supervised training is a means for reducing the effort needed to prepare the training set by training the model with a small number of fully labeled examples and an additional set of unlabeled or weakly labeled examples. In this work we present a semi-supervised approach to training object detection systems based on self-training. We implement our approach as a wrapper around the training process of an existing object detector and present empirical results. The key contributions of this empirical study is to demonstrate that a model trained in this manner can achieve results comparable to a model trained in the traditional manner using a much larger set of fully labeled data, and that a training data selection metric that is defined independently of the detector greatly outperforms a selection metric based on the detection confidence generated by the detector.

A theoretical assessment of the user-satisfaction construct in information systems research

Abstract: For nearly two decades, the user-satisfaction construct has occupied a central role in behavioral research in Information Systems IS. In industry, the construct has often been used as a surrogate for IS effectiveness. Given its widespread use by both academics and practitioners, it is surprising that no comprehensive theoretical assessment of this construct has been performed. This paper provides such a review. It begins by examining conceptual and theoretical limitations of the construct's use as a measure of IS effectiveness. Attention is then focused on the evolution of the construct in the literature and the theoretical problems associated with its broader use. The fundamental similarity between user satisfaction and the social and cognitive psychologists' notion of an attitude is suggested. The next sections focus on a discussion of attitude structures and function. First, alternative theoretical views on attitude structure are presented. While one of these structures, the family of expectancy-value models, is reflected in current research on user satisfaction, the second, the family of cognitive approaches, is not. The two attitude structures are considered from the perspective of possible refinements to future work in IS. Next, an examination is made of the ways in which these structures have been integrated in terms of understanding the relationship of users' affective responses to other responses i.e., behavior or cognition. This leads to a discussion of the function attitudes might serve for the user other than the evaluation of an information system or IS staff. Finally, the question of how behavior influences attitude is considered. The paper concludes with suggestions for future work.

Learning to extract symbolic knowledge from the World Wide Web

Abstract: The World Wide Web is a vast source of information accessible to computers, but understandable only to humans. The goal of the research described here is to automatically create a computer understandable world wide knowledge base whose content mirrors that of the World Wide Web. Such a knowledge base would enable much more effective retrieval of Web information, and promote new uses of the Web to support knowledge-based inference and problem solving. Our approach is to develop a trainable information extraction system that takes two inputs: an ontology defining the classes and relations of interest, and a set of training data consisting of labeled regions of hypertext representing instances of these classes and relations. Given these inputs, the system learns to extract information from other pages and hyperlinks on the Web. This paper describes our general approach, several machine learning algorithms for this task, and promising initial results with a prototype system.

LEAF: A Benchmark for Federated Settings

Abstract: Modern federated networks, such as those comprised of wearable devices, mobile phones, or autonomous vehicles, generate massive amounts of data each day. This wealth of data can help to learn models that can improve the user experience on each device. However, the scale and heterogeneity of federated data presents new challenges in research areas such as federated learning, meta-learning, and multi-task learning. As the machine learning community begins to tackle these challenges, we are at a critical time to ensure that developments made in these areas are grounded with realistic benchmarks. To this end, we propose LEAF, a modular benchmarking framework for learning in federated settings. LEAF includes a suite of open-source federated datasets, a rigorous evaluation framework, and a set of reference implementations, all geared towards capturing the obstacles and intricacies of practical federated environments.

Scaling Factors for Obtaining Fundamental Vibrational Frequencies and Zero-Point Energies from HF/6–31G* and MP2/6–31G* Harmonic Frequencies

Abstract: New scaling factors have been determined for obtaining fundamental vibrational frequencies and zero-point vibrational energies from harmonic frequencies calculated at the HF/6–31G* and MP2/6–31G* levels. The scaling factors for the fundamental frequencies have been derived from a comparison of a total of 1066 calculated frequencies for 122 molecules with corresponding experimental values, while the zero-point energy scaling factors were determined from a comparison of the computed values with the experimental zero-point energies for a set of 24 molecules. The scaling factors recommended are, respectively, 0.8929 and 0.9427 for HF/6–31G* and MP2/6–31G* fundamental frequencies, and 0.9135 and 0.9646 for HF/6–31G* and MP2/6–31G* zero-point energies. RMS errors were determined to be around 50 cm−1 for the HF and MP2 fundamental frequencies, and around 0.4 kJ mol−1 for the HF and MP2 zero-point energies.