Optimal Neural Codes for Control and Estimation

Authors: Alex K. Susemihl, Ron Meir, Manfred Opper

NeurIPS 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The evolution of the average posterior variance is given by the average of equation (6b), which involves nonlinear averages over the covariances. These are intractable, but a simple mean-field approach yields the approximate equation for the evolution of the average Σs = E [Σs|Σ0] ds = A Σs + Σs A + D ˆλ Σs P Σs I + P Σs 1 . The alternative is to simulate the stochastic dynamics of Σt for a large number of samples and compute numerical averages. These results can be directly employed to evaluate the optimal costto-go in the control problem f(Σ, t).
Researcher Affiliation Collaboration Alex Susemihl1, Manfred Opper Methods of Artificial Intelligence Technische Universit at Berlin 1 Current affiliation: Google Ron Meir Department of Electrical Engineering Technion Haifa
Pseudocode No The paper does not contain any pseudocode or clearly labeled algorithm blocks.
Open Source Code No The paper does not provide any statement about releasing open-source code for the described methodology or a link to a code repository.
Open Datasets No The paper conducts theoretical analysis and simulations of stochastic systems (e.g., Ornstein Uhlenbeck process, stochastic damped harmonic oscillator) rather than using a publicly available dataset for training.
Dataset Splits No The paper focuses on theoretical models and simulations of stochastic systems, and as such, it does not specify training, validation, or test dataset splits. The study does not involve empirical evaluation on standard datasets with such splits.
Hardware Specification No The paper does not provide any specific details about the hardware used to run the simulations or conduct the research (e.g., CPU, GPU models, or memory specifications).
Software Dependencies No The paper discusses theoretical frameworks and algorithms like Kalman filtering, LQG control, and point-process filtering, but it does not specify any software dependencies with version numbers (e.g., Python 3.x, specific libraries, or simulation software).
Experiment Setup Yes Parameters for figure (a) were: T = 2, γ = 1.0, η = 0.6, b = 0.2, φ = 0.1, θ = 0.05, Q = 0.1, QT = 0.001, R = 0.1. Parameters for figure (b) were T = 5, γ = 0.4, ω = 0.8, η = 0.4, r = 0.4, q = 0.4, QT = 0, φ = 0.5, θ = 0.1.