Newscast EM
نویسندگان
چکیده
We propose a gossip-based distributed algorithm for Gaussian mixture learning, Newscast EM. The algorithm operates on network topologies where each node observes a local quantity and can communicate with other nodes in an arbitrary point-to-point fashion. The main difference between Newscast EM and the standard EM algorithm is that the M-step in our case is implemented in a decentralized manner: (random) pairs of nodes repeatedly exchange their local parameter estimates and combine them by (weighted) averaging. We provide theoretical evidence and demonstrate experimentally that, under this protocol, nodes converge exponentially fast to the correct estimates in each M-step of the EM algorithm.
منابع مشابه
Multi-Observations Newscast EM for Distributed Multi-Camera Tracking
Visual surveillance in wide areas (e.g. airports) relies on multiple cameras which observe non-overlapping scenes. The focus of this thesis is multi-person tracking, where the task is to maintain a person’s identity when he or she leaves the field of view of one camera and later re-appears at another camera. While current wide-area tracking systems are central systems, we propose to use a distr...
متن کاملMulti-Observations Newscast EM for Distributed Appearance Based Tracking
Visual surveillance in wide areas (e.g. airports) relies on cameras that observe non-overlapping scenes. Multi-person tracking requires re-identification of people, when they leave one field of view and later enter another. For this we use appearance cues. Under the assumption that all observations of a single person are Gaussian distributed, the observation model in our approach consists of a ...
متن کاملA distributed EM algorithm for large peer-to-peer networks
We propose a new distributed algorithm for Gaussian mixture learning: Newscast EM. The algorithm can run on large peer-to-peer networks in which each node observes a local quantity and can communicate with other nodes in a random, anonymous fashion. The main difference with the standard (centralized) EM algorithm is the M-step which is implemented in a decentralized manner: random pairs of node...
متن کامل6.111 Final Project Proposal
Overview Our final project involves a do-it-yourself newscast. It requires a main feed that has a person standing in front of a distinguishable screen and preloaded pictures and corresponding captions. The user is also allowed basic control of the system including changing between different picture and captions and also making the picture become fullscreen to include more detail. This system al...
متن کاملNewscast Computing
Monitoring large computer networks often involves aggregation of various sorts of data that are distributed across network components. Finding extreme values, counting discrete observations or computing an average or a sum of some parameter values are typical examples of such “background” activities that provide input to monitoring systems. Another aspect of network management is fast and relia...
متن کامل