Discovery of activity patterns using topic models

Huỳnh, Tâm; Fritz, Mario; Schiele, Bernt

doi:10.1145/1409635.1409638

Cited by 354 publications

(290 citation statements)

References 16 publications

Supporting

Mentioning

287

Contrasting

Unclassified

Order By: Relevance

“…Another closely related work published simultaneously to ours GaticaPerez [2008a][2008b]], is described in [Huynh et al 2008] and uses topic models for human activity discovery, using wearable sensor data and not mobile phone data. The method identifies activity patterns in one single person's daily life over sixteen days, using two wearable sensors, one placed on the right hip and the other on the right wrist.…”

Section: Related Workmentioning

confidence: 99%

“…Further, a document is constructed from a sliding window of length D. In contrast, our work investigates the human routine discovery task from mobile phone data, on a large scale, and we use this data to discover group routines in addition to individual routines. Our documents are independent and identically distributed as in topic models, though this is not the case in [Huynh et al 2008]. Further, our methodology has proven successful on lower level input data which can be obtained more directly from sensor data, such as the locations of an individual and their proximate interactions [Farrahi and Gatica-Perez [2008a]].…”

Section: Related Workmentioning

confidence: 99%

“…Further, our methodology has proven successful on lower level input data which can be obtained more directly from sensor data, such as the locations of an individual and their proximate interactions [Farrahi and Gatica-Perez [2008a]]. The methodology proposed in [Huynh et al 2008] requires higher level information regarding a person's activities through the use of multiple devices attached to various body parts, unlike our work which only relies on one single device (a phone) which is worn and used naturally. Further, our methodology also investigates the Author-Topic model [Rosen-Zvi et al 2004].…”

Section: Related Workmentioning

confidence: 99%

“…Recently, they have been successfully applied to data sources other than text, such as images Quelhas et al 2007], video [Niebles et al 2006], genetics [Pritchard et al 2000], and wearable sensor data [Huynh et al 2008], but to our knowledge, their use for real-life routine modeling from large-scale mobile phone data is novel. Topic models are generative models that represent documents as mixtures of topics, learned in a latent space, and they allow for clustering and ranking of documents, words, and other entities, like authors [Rosen-Zvi et al 2004].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Discovering routines from large-scale human locations using probabilistic topic models

Farrahi

Gática-Pérez

2011

ACM Trans. Intell. Syst. Technol.

196

109

View full text Add to dashboard Cite

In this work we discover the daily location-driven routines which are contained in a massive reallife human dataset collected by mobile phones. Our goal is the discovery and analysis of human routines which characterize both individual and group behaviors in terms of location patterns. We develop an unsupervised methodology based on two differing probabilistic topic models and apply them to the daily life of 97 mobile phone users over a 16 month period to achieve these goals. Topic models are probabilistic generative models for documents that identify the latent structure that underlies a set of words. Routines dominating the entire group's activities, identified with a methodology based on the Latent Dirichlet Allocation topic model, include "going to work late", "going home early", "working non-stop" and "having no reception (phone off)" at different times over varying time-intervals. We also detect routines which are characteristic of users, with a methodology based on the Author-Topic model. With the routines discovered, and the two methods of characterizing days and users, we can then perform various tasks. We use the routines discovered to determine behavioral patterns of users and groups of users. For example, we can find individuals that display specific daily routines, such as "going to work early" or "turning off the mobile (or having no reception) in the evenings". We are also able to characterize daily patterns by determining the topic structure of days in addition to determining whether certain routines occur dominantly on weekends or weekdays. Furthermore, the routines discovered can be used to rank users or find subgroups of users who display certain routines. We can also characterize users based on their entropy. We compare our method to one based on clustering using K-means. Finally, we analyze an individual's routines over time to determine regions with high variations, which may correspond to specific events.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Discovering routines from large-scale human locations using probabilistic topic models

Farrahi

Gática-Pérez

2011

ACM Trans. Intell. Syst. Technol.

196

109

View full text Add to dashboard Cite

show abstract

“…Other existing works have found typical patterns by clustering human behavior. Indoor daily routines, like commuting and office work, were discovered by Huynh et al [21] using wearable sensors and accelerometer data with applications in elderly care, or office space management. Outdoor daily routines were discovered by Eagle and Pentland [12] using Principal Component Analysis (PCA) from mobile phone data on subjects' location, proximity, communication, and device usage behavior.…”

Section: Human Behavior Analysis Using Mobile Sensorsmentioning

confidence: 99%

Privacy-sensitive recognition of group conversational context with sociometers

et al. 2011

View full text Add to dashboard Cite

Recognizing the conversational context in which group interactions unfold has applications in machines that support collaborative work and perform automatic social inference using contextual knowledge. This paper addresses the task of discriminating one conversational context from another, specifically brainstorming from decision-making interactions, using easily computable nonverbal behavioral cues. Privacy-sensitive mobile sociometers are used to record the interaction data. We hypothesize that the difference in the conversational dynamics between brainstorming and decision-making discussions is significant and measurable using speaking activity-based nonverbal cues. We characterize the communication patterns of the entire group by the aggregation (both temporal and person-wise) of their nonverbal behavior. The results on our interaction data set show that the floor-occupation patterns in a brainstorming interaction are different from a decision-making interaction, and our method can obtain a classification accuracy as high as 87.5%.

show abstract

Bibliography

2015

Activity Learning

View full text Add to dashboard Cite

Discovery of activity patterns using topic models

Cited by 354 publications

References 16 publications

Discovering routines from large-scale human locations using probabilistic topic models

Discovering routines from large-scale human locations using probabilistic topic models

Privacy-sensitive recognition of group conversational context with sociometers

Bibliography

Contact Info

Product

Resources

About