Search results for " Inference"
showing 7 items of 337 documents
Enhancing identification of causal effects by pruning
2018
Causal models communicate our assumptions about causes and effects in real-world phe- nomena. Often the interest lies in the identification of the effect of an action which means deriving an expression from the observed probability distribution for the interventional distribution resulting from the action. In many cases an identifiability algorithm may return a complicated expression that contains variables that are in fact unnecessary. In practice this can lead to additional computational burden and increased bias or inefficiency of estimates when dealing with measurement error or missing data. We present graphical criteria to detect variables which are redundant in identifying causal effe…
Causality-Aware Convolutional Neural Networks for Advanced Image Classification and Generation
2023
Smart manufacturing uses emerging deep learning models, and particularly Convolutional Neural Networks (CNNs) and Generative Adversarial Networks (GANs), for different industrial diagnostics tasks, e.g., classification, detection, recognition, prediction, synthetic data generation, security, etc., on the basis of image data. In spite of being efficient for these objectives, the majority of current deep learning models lack interpretability and explainability. They can discover features hidden within input data together with their mutual co-occurrence. However, they are weak at discovering and making explicit hidden causalities between the features, which could be the reason behind the parti…
Explainable Fuzzy AI Challenge 2022 : Winner’s Approach to a Computationally Efficient and Explainable Solution
2022
An explainable artificial intelligence (XAI) agent is an autonomous agent that uses a fundamental XAI model at its core to perceive its environment and suggests actions to be performed. One of the significant challenges for these XAI agents is performing their operation efficiently, which is governed by the underlying inference and optimization system. Along similar lines, an Explainable Fuzzy AI Challenge (XFC 2022) competition was launched, whose principal objective was to develop a fully autonomous and optimized XAI algorithm that could play the Python arcade game “Asteroid Smasher”. This research first investigates inference models to implement an efficient (XAI) agent using rule-based …
Continuum: A spatiotemporal data model to represent and qualify filiation relationships
2013
International audience; This work introduces an ontology-based spatio-temporal data model to represent entities evolving in space and time. A dynamic phenomenon generates a complex relationship network between the entities involved in the process. At the abstract level, the relationships can be identity or topological filiations. The existence of an identity filiation depends on whether the object changes its identity or not. On the other hand, topological filiations are based exclusively on the spatial component, like in the case of growth, reduction, merging or splitting. When combining identity and topological filiations, six filiation relationships are obtained, forming a second abstrac…
The Spatial Overlap of Police Calls Reporting Street-Level and Behind-Closed-Doors Crime: A Bayesian Modeling Approach
2021
Traditionally, intimate-partner violence has been considered a special type of crime that occurs behind closed doors, with different characteristics from street-level crime. The aim of this study is to analyze the spatial overlap of police calls reporting street-level and behind-closed-doors crime. We analyzed geocoded police calls in the 552 census-block groups of the city of Valencia, Spain, related to street-level crime (N = 26,624) and to intimate-partner violence against women (N = 11,673). A Bayesian joint model was run to analyze the spatial overlap. In addition, two Bayesian hierarchical models controlled for different neighborhood characteristics to analyze the relative risks. Resu…
‘Seeing the Dark’: Grounding Phenomenal Transparency and Opacity in Precision Estimation for Active Inference
2018
One of the central claims of the Self-model Theory of Subjectivity is that the experience of being someone - even in a minimal form - arises through a transparent phenomenal self-model, which itself can in principle be reduced to brain processes. Here, we consider whether it is possible to distinguish between phenomenally transparent and opaque states in terms of active inference. We propose a relationship of phenomenal opacity to expected uncertainty or precision; i.e., the capacity for introspective attention and implicit mental action. Thus we associate introspective attention with the deployment of 'precision' that may render the perceptual evidence (for action) opaque, while treating t…
Prediction of the next value of a function
1981
The following model of inductive inference is considered. Arbitrary set tau = {tau_1, tau_2, ..., tau_n} of n total functions N->N is fixed. A "black box" outputs the values f(0), f(1), ..., f(m), ... of some function f from the set tau. Processing these values by some algorithm (a strategy) we try to predict f(m+1) from f(0), f(1), ..., f(m). Upper and lower bounds for average error numbers are obtained for prediction by using deterministic and probabilistic strategies.