PROPS: Probabilistic personalization of black-box sequence models

@article{Wojnowicz2018PROPSPP,
  title={PROPS: Probabilistic personalization of black-box sequence models},
  author={Michael Thomas Wojnowicz and Xuan Zhao},
  journal={2018 IEEE International Conference on Big Data (Big Data)},
  year={2018},
  pages={4768-4774}
}
  • M. Wojnowicz, Xuan Zhao
  • Published 1 December 2018
  • Computer Science
  • 2018 IEEE International Conference on Big Data (Big Data)
We present PROPS, a lightweight transfer learning mechanism for sequential data. PROPS learns probabilistic perturbations around the predictions of one or more arbitrarily complex, pre-trained black box models (such as recurrent neural networks). The technique pins the black-box prediction functions to "source states" of a hidden Markov model (HMM), and uses the remaining states as "perturbation states" for learning customized perturbations around those predictions. In this paper, we describe… 

Figures and Tables from this paper

References

SHOWING 1-9 OF 9 REFERENCES
Efficient Transfer Learning Schemes for Personalized Language Modeling using Recurrent Neural Network
TLDR
Through experiments on dialogue data in a drama, it is verified that the proposed transfer learning methods have successfully generated the personalized language model, whose output is more similar to the personal language style in both qualitative and quantitative aspects.
Comparing Hidden Markov Models and Long Short Term Memory Neural Networks for Learning Action Representations
TLDR
It is shown that, despite the current trend towards (deep) neural networks, traditional graphical model approaches are still beneficial under conditions where only few data points or limited computing power is available.
Sequential Neural Models with Stochastic Layers
TLDR
Stochastic recurrent neural networks are introduced which glue a deterministic recurrent neural network and a state space model together to form a stochastic and sequential neural generative model.
Long Short-Term Memory
TLDR
A novel, efficient, gradient based method called long short-term memory (LSTM) is introduced, which can learn to bridge minimal time lags in excess of 1000 discrete-time steps by enforcing constant error flow through constant error carousels within special units.
Online EM Algorithm for Hidden Markov Models
TLDR
Although the proposed online EM algorithm resembles a classical stochastic approximation (or Robbins–Monro) algorithm, it is sufficiently different to resist conventional analysis of convergence and provides limited results which identify the potential limiting points of the recursion as well as the large-sample behavior of the quantities involved in the algorithm.
A survey of techniques for incremental learning of HMM parameters
A probabilistic distance measure for hidden Markov models
We propose a probabilistic distance measure for measuring the dissimilarity between pairs of hidden Markov models with arbitrary observation densities. The measure is based on the Kullback-Leibler
CONSISTENCY AND IDENTIFIABILITY REVISITED
Summary We provide a general framework to review the well-known concept of identifiability and give a formal proof that this is implied by the existence of a consistent estimator. We apply these
CONSISTENCY OF THE MAXIMUM LIKELIHOOD ESTIMATOR FOR GENERAL HIDDEN MARKOV MODELS
TLDR
It is proved that the maximum likelihood estimator (MLE) of the parameter is strongly consistent under a rather minimal set of assumptions, which could form a foundation for the investigation of MLE consistency in more general dependent and non-Markovian time series.