Chapter 17. Logic Tensor Networks: Theory and Applications

2021 ◽  
Author(s):  
Luciano Serafini ◽  
Artur d’Avila Garcez ◽  
Samy Badreddine ◽  
Ivan Donadello ◽  
Michael Spranger ◽  
...  
Keyword(s):  
Supervised Learning ◽  
Data Sets ◽  
Symbolic Representations ◽  
Tensor Networks ◽  
Multiple Data ◽  
Abstract Knowledge ◽  
Symbolic Ai ◽  
Rich Data ◽  
Efficient Learning ◽  
High Level

The recent availability of large-scale data combining multiple data modalities has opened various research and commercial opportunities in Artificial Intelligence (AI). Machine Learning (ML) has achieved important results in this area mostly by adopting a sub-symbolic distributed representation. It is generally accepted now that such purely sub-symbolic approaches can be data inefficient and struggle at extrapolation and reasoning. By contrast, symbolic AI is based on rich, high-level representations ideally based on human-readable symbols. Despite being more explainable and having success at reasoning, symbolic AI usually struggles when faced with incomplete knowledge or inaccurate, large data sets and combinatorial knowledge. Neurosymbolic AI attempts to benefit from the strengths of both approaches combining reasoning with complex representation of knowledge and efficient learning from multiple data modalities. Hence, neurosymbolic AI seeks to ground rich knowledge into efficient sub-symbolic representations and to explain sub-symbolic representations and deep learning by offering high-level symbolic descriptions for such learning systems. Logic Tensor Networks (LTN) are a neurosymbolic AI system for querying, learning and reasoning with rich data and abstract knowledge. LTN introduces Real Logic, a fully differentiable first-order language with concrete semantics such that every symbolic expression has an interpretation that is grounded onto real numbers in the domain. In particular, LTN converts Real Logic formulas into computational graphs that enable gradient-based optimization. This chapter presents the LTN framework and illustrates its use on knowledge completion tasks to ground the relational predicates (symbols) into a concrete interpretation (vectors and tensors). It then investigates the use of LTN on semi-supervised learning, learning of embeddings and reasoning. LTN has been applied recently to many important AI tasks, including semantic image interpretation, ontology learning and reasoning, and reinforcement learning, which use LTN for supervised classification, data clustering, semi-supervised learning, embedding learning, reasoning and query answering. The chapter presents some of the main recent applications of LTN before analyzing results in the context of related work and discussing the next steps for neurosymbolic AI and LTN-based AI models.

scholarly journals Semi-Supervised Learning Using Hierarchical Mixture Models: Gene Essentiality Case Study

2021 ◽  
Vol 26 (2) ◽  
pp. 40
Author(s):  
Michael W. Daniels ◽  
Daniel Dvorkin ◽  
Rani K. Powers ◽  
Katerina Kechris
Keyword(s):  
Supervised Learning ◽  
Mixture Model ◽  
Essential Genes ◽  
Training Data ◽  
Data Sets ◽  
Data Types ◽  
Gene Essentiality ◽  
Multiple Data ◽  
Level Data ◽  
Accuracy Of Prediction

Integrating gene-level data is useful for predicting the role of genes in biological processes. This problem has typically focused on supervised classification, which requires large training sets of positive and negative examples. However, training data sets that are too small for supervised approaches can still provide valuable information. We describe a hierarchical mixture model that uses limited positively labeled gene training data for semi-supervised learning. We focus on the problem of predicting essential genes, where a gene is required for the survival of an organism under particular conditions. We applied cross-validation and found that the inclusion of positively labeled samples in a semi-supervised learning framework with the hierarchical mixture model improves the detection of essential genes compared to unsupervised, supervised, and other semi-supervised approaches. There was also improved prediction performance when genes are incorrectly assumed to be non-essential. Our comparisons indicate that the incorporation of even small amounts of existing knowledge improves the accuracy of prediction and decreases variability in predictions. Although we focused on gene essentiality, the hierarchical mixture model and semi-supervised framework is standard for problems focused on prediction of genes or other features, with multiple data types characterizing the feature, and a small set of positive labels.

Multiple data parameter identification for nonlinear conceptual models

1997 ◽  
Vol 36 (5) ◽  
pp. 61-68 ◽  
Author(s):  
Hermann Eberl ◽  
Amar Khelil ◽  
Peter Wilderer
Keyword(s):  
Optimization Problem ◽  
Conceptual Models ◽  
Reference Data ◽  
Transport Model ◽  
Calibration Method ◽  
Data Sets ◽  
Multiple Data ◽  
Marquardt Algorithm ◽  
Multicriteria Optimization Problem ◽  
Higher Order Differential Equations

A numerical method for the identification of parameters of nonlinear higher order differential equations is presented, which is based on the Levenberg-Marquardt algorithm. The estimation of the parameters can be performed by using several reference data sets simultaneously. This leads to a multicriteria optimization problem, which will be treated by using the Pareto optimality concept. In this paper, the emphasis is put on the presentation of the calibration method. As an example identification of the parameters of a nonlinear hydrological transport model for urban runoff is included, but the method can be applied to other problems as well.

scholarly journals Automatic Labeled Dialogue Generation for Nursing Record Systems

2020 ◽  
Vol 10 (3) ◽  
pp. 62
Author(s):  
Tittaya Mairittha ◽  
Nattaya Mairittha ◽  
Sozo Inoue
Keyword(s):  
Data Augmentation ◽  
Short Term Memory ◽  
Generative Models ◽  
Abstract Knowledge ◽  
Augmentation Techniques ◽  
Nursing Record ◽  
Long Short Term Memory ◽  
The Individual ◽  
High Level ◽  
Embedding Methods

The integration of digital voice assistants in nursing residences is becoming increasingly important to facilitate nursing productivity with documentation. A key idea behind this system is training natural language understanding (NLU) modules that enable the machine to classify the purpose of the user utterance (intent) and extract pieces of valuable information present in the utterance (entity). One of the main obstacles when creating robust NLU is the lack of sufficient labeled data, which generally relies on human labeling. This process is cost-intensive and time-consuming, particularly in the high-level nursing care domain, which requires abstract knowledge. In this paper, we propose an automatic dialogue labeling framework of NLU tasks, specifically for nursing record systems. First, we apply data augmentation techniques to create a collection of variant sample utterances. The individual evaluation result strongly shows a stratification rate, with regard to both fluency and accuracy in utterances. We also investigate the possibility of applying deep generative models for our augmented dataset. The preliminary character-based model based on long short-term memory (LSTM) obtains an accuracy of 90% and generates various reasonable texts with BLEU scores of 0.76. Secondly, we introduce an idea for intent and entity labeling by using feature embeddings and semantic similarity-based clustering. We also empirically evaluate different embedding methods for learning good representations that are most suitable to use with our data and clustering tasks. Experimental results show that fastText embeddings produce strong performances both for intent labeling and on entity labeling, which achieves an accuracy level of 0.79 and 0.78 f1-scores and 0.67 and 0.61 silhouette scores, respectively.

The development of nurses’ foundational values

2021 ◽  
pp. 096973302110032
Author(s):  
Sastrawan Sastrawan ◽  
Jennifer Weller-Newton ◽  
Gabrielle Brand ◽  
Gulzar Malik
Keyword(s):  
Professional Training ◽  
Reference Points ◽  
Constructivist Grounded Theory ◽  
Professional Values ◽  
Data Sets ◽  
Value System ◽  
Institutional Values ◽  
Multiple Data ◽  
Human Ethics ◽  
Multiple Data Sets

Background: In the ever-changing and complex healthcare environment, nurses encounter challenging situations that may involve a clash between their personal and professional values resulting in a profound impact on their practice. Nevertheless, there is a dearth of literature on how nurses develop their personal–professional values. Aim: The aim of this study was to understand how nurses develop their foundational values as the base for their value system. Research design: A constructivist grounded theory methodology was employed to collect multiple data sets, including face-to-face focus group and individual interviews, along with anecdote and reflective stories. Participants and research context: Fifty-four nurses working across various nursing settings in Indonesia were recruited to participate. Ethical considerations: Ethics approval was obtained from the Monash University Human Ethics Committee, project approval number 1553. Findings: Foundational values acquisition was achieved through family upbringing, professional nurse education and organisational/institutional values reinforcement. These values are framed through three reference points: religious lens, humanity perspective and professionalism. This framing results in a unique combination of personal–professional values that comprise nurses’ values system. Values are transferred to other nurses either in a formal or informal way as part of one’s professional responsibility and customary social interaction via telling and sharing in person or through social media. Discussion: Values and ethics are inherently interweaved during nursing practice. Ethical and moral values are part of professional training, but other values are often buried in a hidden curriculum, and attained and activated through interactions during nurses’ training. Conclusion: Developing a value system is a complex undertaking that involves basic social processes of attaining, enacting and socialising values. These processes encompass several intertwined entities such as the sources of values, the pool of foundational values, value perspectives and framings, initial value structures, and methods of value transference.

scholarly journals The role of cultural practices in the emergence of modern human intelligence

2008 ◽  
Vol 363 (1499) ◽  
pp. 2011-2019 ◽  
Author(s):  
Edwin Hutchins
Keyword(s):  
Embodied Cognition ◽  
Cognitive Processes ◽  
Human Activities ◽  
Cultural Practices ◽  
Modern Human ◽  
Human Intelligence ◽  
Symbolic Representations ◽  
High Level ◽  
Human Brains

Innate cognitive capacities are orchestrated by cultural practices to produce high-level cognitive processes. In human activities, examples of this phenomenon range from everyday inferences about space and time to the most sophisticated reasoning in scientific laboratories. A case is examined in which chimpanzees enter into cultural practices with humans (in experiments) in ways that appear to enable them to engage in symbol-mediated thought. Combining the cultural practices perspective with the theories of embodied cognition and enactment suggests that the chimpanzees' behaviour is actually mediated by non-symbolic representations. The possibility that non-human primates can engage in cultural practices that give them the appearance of symbol-mediated thought opens new avenues for thinking about the coevolution of human culture and human brains.

scholarly journals Evaluation of daily precipitation statistics and monsoon onset/retreat over western Sahel in multiple data sets

2014 ◽  
Vol 45 (5-6) ◽  
pp. 1325-1354 ◽  
Author(s):  
Emilia Paula Diaconescu ◽  
Philippe Gachon ◽  
John Scinocca ◽  
René Laprise
Keyword(s):  
Daily Precipitation ◽  
Monsoon Onset ◽  
Data Sets ◽  
Multiple Data ◽  
Western Sahel ◽  
Multiple Data Sets

Can earning much lead to doing good? Firm performance implications for corporate social responsibility

2021 ◽  
Vol ahead-of-print (ahead-of-print) ◽  
Author(s):  
Siyu Hou ◽  
Zhaoyang Guo ◽  
Chuangneng Cai ◽  
Xiaobo Jiao
Keyword(s):  
Corporate Social Responsibility ◽  
Firm Performance ◽  
Social Responsibility ◽  
State Ownership ◽  
Data Sets ◽  
Content Type ◽  
Multiple Data ◽  
Instrumental Variable Regression ◽  
Multiple Data Sets ◽  
Corporate Social

Purpose The purpose of this study is to examine the influence of firm performance on corporate social responsibility (CSR) and its possible moderating effect. Despite the significance of CSR, there remains an extensive debate about how it is affected by firm performance. Design/methodology/approach The conceptual model is mainly built on goal-setting theory. Based on archival data from multiple data sets on 1,650 companies, collected from 2010 to 2017, the hypotheses are tested using the two-stage instrumental variable regression method. Findings There is an inverted U-shaped relationship between firm performance and CSR that first increases and then decreases. In addition, considering the boundary conditions, state ownership makes the inverted U-shaped curve steeper, while high executive wage concentration makes the inverted U-shaped curve flatter. Research limitations/implications This study harmonizes the traditional contradictory findings of the influence of firm performance on CSR, that is, it supports a positive, negative or neutral relationship between the two. Originality/value This research provides a necessary structure for the CSR literature. By delving deeply into the relationship between firm performance and CSR, it enables scholars to better address the critical management question of whether earning more will lead to doing good.

Sign in / Sign up

Export Citation Format

Share Document