{
  "id": "1611.05545",
  "version": "v1",
  "published": "2016-11-17T03:02:01.000Z",
  "updated": "2016-11-17T03:02:01.000Z",
  "title": "Stochastic Gradient Descent in Continuous Time",
  "authors": [
    "Justin Sirignano",
    "Konstantinos Spiliopoulos"
  ],
  "categories": [
    "math.PR",
    "math.OC",
    "math.ST",
    "stat.ML",
    "stat.TH"
  ],
  "abstract": "We consider stochastic gradient descent for continuous-time models. Traditional approaches for the statistical estimation of continuous-time models, such as batch optimization, can be impractical for large datasets where observations occur over a long period of time. Stochastic gradient descent provides a computationally efficient method for such statistical estimation problems. The stochastic gradient descent algorithm performs an online parameter update in continuous time, with the parameter updates satisfying a stochastic differential equation. The parameters are proven to converge to a local minimum of a natural objective function for the estimation of the continuous-time dynamics. The convergence proof leverages ergodicity by using an appropriate Poisson equation to help describe the evolution of the parameters for large times. Numerical analysis of the stochastic gradient descent algorithm is presented for several examples, including the Ornstein-Uhlenbeck process, Burger's stochastic partial differential equation, and reinforcement learning.",
  "revisions": [
    {
      "version": "v1",
      "updated": "2016-11-17T03:02:01.000Z"
    }
  ],
  "analyses": {
    "keywords": [
      "continuous time",
      "burgers stochastic partial differential equation",
      "stochastic gradient descent algorithm performs",
      "convergence proof leverages ergodicity"
    ],
    "note": {
      "typesetting": "TeX",
      "pages": 0,
      "language": "en",
      "license": "arXiv",
      "status": "editable"
    }
  }
}