Learning Influence among Interacting Markov Chains

By Yoshua Bengio, Paolo Frasconi. 2005

260 citations recorded by the source corpus.

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

Hilbert Space Embeddings of Predictive State Representations (2013, methodology): “If the observations are discrete, then Hidden Markov Models (HMMs) [1] or, in the control setting, Input-Output HMMs (IOHMMs) [2], can be used to represent belief as a discrete distribution over latent states.”
The Hierarchical Hidden Markov Model: Analysis and Applications (1998, methodology): “For instance, using the framework introduced by Bengio and Frasconi (1995) for input-output HMMs, hierarchical HMMs can be generalized to describe input-output mappings between strings over two different alphabets.”
Learning predictive models of a depth camera & manipulator from raw execution traces (2014, methodology): “We dispense with problem-dependent geometric and physical intuitions and instead model the sensorimotor data as a Predictive State Representation (PSR) [6], [7], a general probabilistic modeling framework that can represent a wide variety of stochastic process models including Kalman filters (KFs) [”
Modeling and forecasting electricity prices with input/output hidden Markov models (2005, background): “IOHMMs appeared in 1995 [23] as an attempt to solve the main drawbacks of these last two models.”
Sensor faults: Detection methods and prevalence in real-world datasets (2010, methodology): “Similarly, we can enhance our HMM-based method to take measurements from other sensors as input [Bengio and Frasconi 1995].”
POPCORN: Partially Observed Prediction COnstrained ReiNforcement Learning (2020, background): “(5) This IO-HMM likelihood marginalizes over uncertainty about the hidden states, can be computed efficiently via dynamic programming (Rabiner, 1989), and is also amenable to automatic differentiation w.r.t. ✓.”
Temporal Models for Predicting Student Dropout in Massive Open Online Courses (2015, methodology): “Recognizing this nature of the problem can help to justify our approach of adopting temporal models, which include the input-output hidden Markov model (IOHMM) [5] and the recurrent neural network (RNN) model [6].”
A Tutorial on Energy-Based Learning (2006, methodology): “Finally, it is the loss function of choice for training other probabilistic discriminative sequence labeling models such as input/output HMM [10], conditional random fields [40], and discriminative random fields [39].”
Offline policy evaluation across representations with applications to educational games (2014, methodology): “We learn IOHMM/POMDP models (Section 7.2) with various feature compaction methodologies (Section 7.1), and two features were selected by hand which were predictive of quitting in other tasks.”
Hierarchical Convolutional Deep Learning in Computer Vision (2013, methodology): “Among probabilistic models, the Input-Output HMM [6] is most similar to the architecture we propose.”