Proceedings ArticleDOI
Uncertainty in neural networks
P.A. Ligomenides
- pp 83-89
TLDR
The authors suggest that uncertainty may be managed naturally, and even used profitably, in cooperative, self-organizing, dynamical physical systems, and in neural networks.Abstract:
Uncertainty in AI applications, as they apply to inductive inference, is often dealt with by modeling heuristic methods of inference based on different kinds of logic, binary, multivalued or fuzzy, simulated on digital computers with probability, possibility or belief theories. The authors suggest that uncertainty may be managed naturally, and even used profitably, in cooperative, self-organizing, dynamical physical systems, and in neural networks. New classes of powerful cooperative computation and learning (C&L) machines are possible, with the class of artificial neural networks being just an early, rather rudimentary, example. The temporal behavior of classical dynamical physical systems was investigated for C&L models. The case of deterministic chaos is considered in studying C&L properties in dynamical system behavior. Deterministic chaos underlines the behavior of a class of physical systems of special interest, whose unpredictability is derived from sensitive dependence on initial conditions in a sustained way which exaggerates uncertainty. Non-Lipschitzian unpredictability is considered. >read more
Citations
More filters
Posted Content
Depth Uncertainty in Neural Networks
TL;DR: This work performs probabilistic reasoning over the depth of neural networks to exploit the sequential structure of feed-forward networks and provide uncertainty calibration, robustness to dataset shift, and accuracies competitive with more computationally expensive baselines.
Journal ArticleDOI
Molecule Identification with Rotational Spectroscopy and Probabilistic Deep Learning.
TL;DR: A proof-of-concept framework for identifying molecules of unknown elemental composition and structure using experimental rotational data and probabilistic deep learning is presented, and four neural network architectures that yield information to assist in the identification of an unknown molecule are described.
Journal ArticleDOI
Bayesian inference of metabolic kinetics from genome-scale multiomics data
TL;DR: It is demonstrated that detailed information on the posterior distribution of parameters can be obtained efficiently at a variety of problem scales, including nearly genome-scale kinetic models trained on multiomics datasets and Bayesian machine learning tools to be leveraged in understanding biological data and in developing new, efficient strain designs.
Posted Content
Inspect, Understand, Overcome: A Survey of Practical Methods for AI Safety.
Sebastian Houben,Stephanie Abrecht,Maram Akila,Andreas Bär,Felix Brockherde,Patrick Feifel,Tim Fingscheidt,Sujan Sai Gannamaneni,Seyed Eghbal Ghobadi,Ahmed Hammam,Anselm Haselhoff,Felix Hauser,Christian Heinzemann,Marco Hoffmann,Nikhil Kapoor,Falk Kappel,Marvin Klingner,Jan Kronenberger,Fabian Küppers,Jonas Löhdefink,Michael Mlynarski,Michael Mock,Firas Mualla,Svetlana Pavlitskaya,Maximilian Poretschkin,Alexander Pohl,Varun Ravi Kumar,Julia Rosenzweig,Matthias Rottmann,Stefan Rüping,Timo Sämann,Jan David Schneider,Elena Schulz,Gesina Schwalbe,Joachim Sicking,Toshika Srivastava,Serin Varghese,Michael Weber,Sebastian J. Wirkert,Tim Wirtz,Matthias Woehrle +40 more
TL;DR: In this article, the authors provide a structured and broad overview of the state-of-the-art techniques aiming to address the model-inherent shortcomings of deep neural networks (DNNs).
Perceptual Attention-based Predictive Control.
TL;DR: The proposed architecture is the first step towards using deep learning-based perceptual control policies in safety-critical domains and outperforms previous approaches on early detection of unsafe conditions, such as when novel obstacles are present in the navigation environment.
References
More filters
Journal ArticleDOI
On the nature of turbulence
David Ruelle,Floris Takens +1 more
TL;DR: A mechanism for the generation of turbulence and related phenomena in dissipative systems is proposed in this article, where the authors propose a mechanism for generating turbulence in a dissipative system with respect to dissipative energy.
Journal ArticleDOI
Strange Attractors, Chaotic Behavior, and Information Flow
TL;DR: In this article, it is argued that a physical implementation of such equations is capable of acting as an information source, bringing into the macroscopic variables information not implicit in initial conditions.
Journal ArticleDOI
Terminal attractors in neural networks
TL;DR: It will be shown that terminal attractors can be incorporated into neural networks such that any desired set of these attractors with prescribed basins is provided by an appropriate selection of the synaptic weights.