A Hybrid Discriminative/Generative Approach for Modeling Human Activities

By Jonathan Lester, Tanzeem Choudhury, Nicky Kern, Gaetano Borriello, B. Hannaford. International Joint Conference on Artificial Intelligence, 2005

493 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/8262bd0fb41956f136dd2bc04ff4863584a43ce7

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

An artificial neural network to estimate physical activity energy expenditure and identify physical activity type from an accelerometer. (2009, methodology): “Lester and colleagues (14, 15) reported a similar rate of activity type identiﬁcation from a three-dimensional accelerometer worn at any one of three locations using static classi-ﬁers coupled with a temporally smoothing HMM to estimate activity type.”
Management of probabilistic data: foundations and challenges (2007, background): “This has been studied in the literature: it is #P-complete [75] and DNF formulas admit a FPTRAS (fully poly-time randomized approximation scheme) [55].”
Investigating statistical machine learning as a tool for software development (2008, background): “Advances in low-cost sensing have prompted work on activity modeling and location-aware computing to enhance the limited input and output capabilities of mobile devices [17, 19].”
Extracting Places and Activities from GPS Traces Using Hierarchical Conditional Random Fields (2007, methodology): “We are currently adding more types of sensors to our model, including data collected by a wearable multi-sensor board (Lester et al. 2005; Raj et al. 2006).”
A tutorial on human activity recognition using body-worn inertial sensors (2014, methodology): “Introduced in machine learning and computer vision [Friedman et al. 2000; Torralba et al. 2007], boosting as a variant of decision fusion has been successfully applied to activity recognition as well [Lester et al. 2005; Blanke and Schiele 2009; Zinnen et al. 2009b].”
Mobility Detection Using Everyday GSM Traces (2006, methodology): “that can identify several physical activities including detecting subtle distinctions such as walking on level ground versus up stairs [17].”
Learning to Recognize Daily Actions Using Gaze (2012, unknown): “[16].”
Analyzing features for activity recognition (2005, methodology): “For our experiments, we used data recorded by Intel Research, Seattle [7].”
USC-HAD: a daily activity dataset for ubiquitous activity recognition using wearable sensors (2012, background): “In terms of performance, the study carried out in [14] has shown that the information gained from multi-modal sensors can offset the information lost when activity data is collected from a single location.”
Temporal segmentation and activity classification from first-person sensing (2009, methodology): “[14] use discriminative classifiers and HMMs to recognize a small set of ten actions (e.”
Physical Human Activity Recognition Using Wearable Sensors (2015, background): “It shows a great potential for activity classiﬁcation problems.”
Fine-grained activity recognition by aggregating abstract object usage (2005, background): “This includes unimodal evaluations of activity and social context from audio [3, 19, 18], video [5], and accelerometers [10], multi-modal sensor evaluations [12] and work which optimizes sensor selection for arbitrary activity recognition [2].”
Transfer Learning for Activity Recognition: A Survey (2013, methodology): “such as hidden Markov models and dynamic Bayesian networks have been used to model activity sequences and to smooth recognition results of an ensemble classifier [35].”
A Practical Approach to Recognizing Physical Activities (2006, methodology): “In section 3, we briefly describe the algorithm developed in [7] that automatically picks the most useful features and learns classifiers from these.”
Conditional random fields for activity recognition (2007, methodology): “used a hybrid model to avoid this issue and obtain the benefits of discriminative training as well as the temporal smoothing that comes from using an HMM [63].”
Classification Accuracies of Physical Activities Using Smartphone Motion Sensors (2012, methodology): “Researchers have explored different accelerometer-equipped monitoring devices, such as customized sensor boards [4,5], Actigraph accelerometer [6,7], DynaPort [8], and Pegasus activity monitors [9], for detecting activities.”