Filters








7,265 Hits in 4.3 sec

Forrester's paradox using typicality

Julian Chingoma, Thomas Meyer
2019 South African Forum for Artificial Intelligence Research  
This motivates the enrichment of deontic logic with a notion of typicality which is based on defeasibility, with defeasibility allowing for reasoning about exceptions.  ...  In deontic logic research, it is common for systems to be validated with the use of deontic paradoxes [18] .  ...  Another reason we look at the paradox is that it provides difficulties that the straightforward examples would not, as it has been a challenge for deontic logic researchers [10, 12, 18] .  ... 
dblp:conf/fair2/ChingomaM19 fatcat:gmmpwuthevgobei5xtth2rgiam

Activation gap generators in neural networks

Marelie H. Davel
2019 South African Forum for Artificial Intelligence Research  
In fact, this question has not been addressed for some of the least complicated of neural network architectures: fully-connected feedforward networks with ReLU activations and a limited number of hidden  ...  Acknowledgements This work was partially supported by the National Research Foundation (NRF, Grant Number 109243).  ...  For some samples, activation values would be too high, for others too low.  ... 
dblp:conf/fair2/Davel19 fatcat:lmhfbgo5fbavjeb7ugxbwhfzaa

Neural speech synthesis for resource-scarce languages

Johannes A. Louw
2019 South African Forum for Artificial Intelligence Research  
The proposed acoustic model provides for an efficient implementation, with faster than real time synthesis.  ...  We compare traditional hidden Markov model (HMM)-based acoustic modelling for speech synthesis with the proposed architecture using the World and LPCNet vocoders, giving both objective and MUSHRA based  ...  The tonal languages in South Africa represent a significant challenge for speech synthesis, especially given the resource-scarce environment.  ... 
dblp:conf/fair2/Louw19 fatcat:2i5sdf5ppbaexedzjgsgerosem

Autoencoding variational Bayes for latent Dirichlet allocation

Zach Wolpe, Alta de Waal
2019 South African Forum for Artificial Intelligence Research  
An alternative approach to sampling and optimisation for approximation is a direct mapping between the data and posterior distribution.  ...  All relevant code is well documented and available here: https://www.zachwolpe.com/research.  ...  Experiments A number of experiments were conducted in aid of answering the follow research questions: 1.  ... 
dblp:conf/fair2/WolpeW19 fatcat:pw2alommoba7fgth5hegsaqhei

Automatic detection of abusive South African tweets using a semi-supervised learning approach

Oluwafemi Oriola, Eduan Kotzé
2019 South African Forum for Artificial Intelligence Research  
Major setbacks for detection of abusive South African tweets are inadequacy of annotated corpus and high cost of annotation, which semi-supervised learning solves.  ...  Chi-square statistics is used for the feature selection, while k-means algorithm is used for clustering of data points. By majority voting rule, reliable labels are assigned to the data points.  ...  Proposed method In this section, we formalise the approach used to detect South African abusive tweets.  ... 
dblp:conf/fair2/OriolaK19 fatcat:rl4c46d6avb2lpodtqfsycv22u

Comparison of clustering techniques for residential load profiles in South Africa

Wiebke Toussaint, Deshendran Moodley
2019 South African Forum for Artificial Intelligence Research  
This work compares techniques for clustering metered residential energy consumption data to construct representative daily load profiles in South Africa.  ...  a large, heterogeneous dataset of South African residential energy consumers.  ...  The South African Domestic Electrical Load Study (DELS) Data This section provides an overview and descriptive statistics of the South African Domestic Electrical Load Study (DELS) datasets and details  ... 
dblp:conf/fair2/ToussaintM19 fatcat:m2zfgeyrtngvnnvjzgz6tnncea

Supervised learning and image processing for efficient malaria detection

Michael White, Patrick Marais
2019 South African Forum for Artificial Intelligence Research  
African countries are disproportionately affected, with 92% of global infections and 93% of global deaths falling in the World Health Organisation African region.  ...  The data that was made available for this research was only classified in two classes: as parasitised or non-parasitised.  ... 
dblp:conf/fair2/WhiteM19 fatcat:es6oa3cgbjetjhcvwy7icdij4e

Effective graph sampling of a nonlinear image transform

Mark de Lancey, Inger Fabris-Rotelli
2019 South African Forum for Artificial Intelligence Research  
This algorithm has been shown to have important applications in artificial intelligence and pattern recognition.  ...  Converting graph signals to the spectral domain can also be a costly overhead, and so methods of estimation for filter banks are examined, as well as the design of a good filter bank that may be reused  ...  Feature and texture extraction has important applications in artificial intelligence, pattern recognition and computer vision [4] .  ... 
dblp:conf/fair2/LanceyF19 fatcat:xxycbwxet5csfjnz2w2v44szly

Insights regarding overfitting on noise in deep learning

Marthinus W. Theunissen, Marelie H. Davel, Etienne Barnard
2019 South African Forum for Artificial Intelligence Research  
for.  ...  The main insights are that deep learning models are optimized for training data modularly, with different regions in the function space dedicated to fitting distinct kinds of sample information.  ...  Acknowledgements This work was partially supported by the National Research Foundation (NRF, Grant Number 109243).  ... 
dblp:conf/fair2/TheunissenDB19 fatcat:ek6y2lx4gbgshazcttwsvc5emu

Directed curiosity-driven exploration in hard exploration, sparse reward environments

Asad Jeewa, Anban Pillay, Edgar Jembere
2019 South African Forum for Artificial Intelligence Research  
Training agents in hard exploration, sparse reward environments is a difficult task since the reward feedback is insufficient for meaningful learning.  ...  This requires further research into "intelligent exploration", through hybridising different shaped reward signals and exploration strategies.  ...  Classic work in [6, 13] investigated balancing exploration and exploitation in polynomial time and has inspired much research in the area of intelligent exploration.  ... 
dblp:conf/fair2/JeewaPJ19 fatcat:aescarhzrbgdroe5w7wormoh6a

Modelling uncertain adaptive decisions: Application to KwaZulu-Natal sugarcane growers

C. S. Price, Deshendran Moodley, Anban W. Pillay
2019 South African Forum for Artificial Intelligence Research  
The model was validated using Pitchforth and Mengersen (2013)'s framework for validating expert elicited Bayesian networks.  ...  This model can be used to simulate the cognitive mechanism for a grower agent in a simulation of a sugarcane supply chain.  ...  University Capacity Development Programme (UCDP) for funding the upgrade to Hugin 8.6 and writing retreats; and College of Law and Management Studies, UKZN, for funding transport to the mill.  ... 
dblp:conf/fair2/PriceMP19 fatcat:ddlimsk5y5cfzlkkxb2de6klqe

Input parameter ranking for neural networks in a space weather regression problem

Stefan Lotz, Jacques P. Beukes, Marelie H. Davel
2019 South African Forum for Artificial Intelligence Research  
In this age of rapidly increasing machine learning capability researchers and domain experts need to be cognisant of the dangers of well performing, but un-explainable models.  ...  The second pair of columns shows the ranking for onset phase.  ... 
dblp:conf/fair2/LotzBD19 fatcat:ydpaura4cnck3nq5o4vfdikf4y

Towards a visual framework for the incorporation of knowledge in the phases of machine learning

Conrad Johann Swanepoel, Katherine Mary Malan
2019 South African Forum for Artificial Intelligence Research  
Introduction Machine learning as a component of artificial intelligence, and especially deep learning, has experienced phenomenal growth over the last couple of years.  ...  (for example, the lack of edge cases).  ... 
dblp:conf/fair2/SwanepoelM19 fatcat:m2xiu26kh5b53krthj5d7bjyxe

Evaluation of combined bi-directional branching entropy language models for morphological segmentation of isiXhosa

Lulamile Mzamo, Albert Helberg, Sonja Bosch
2019 South African Forum for Artificial Intelligence Research  
An evaluation of the IsiXhosa Branching Entropy Segmenter (XBES), an unsupervised morphological segmenter for isiXhosa, is presented.  ...  The segmenter contributes a combined bi-directional branching entropy language model with an option for modified Kneser-Ney (mKN) smoothing.  ...  The authors thank the South African Centre for Digital Language Resources (SADiLaR) (https://www.sadilar.org) for providing a central source of data and resources for South African Natural Language Processing  ... 
dblp:conf/fair2/MzamoHB19 fatcat:5ucljxq5bbfjzbqh456p5zhjjm

A cross-comparison of feature selection algorithms on multiple cyber security data-sets

Alexander Powell, Darren Bates, Chad van Wyk, Darren de Abreu
2019 South African Forum for Artificial Intelligence Research  
This paper aims to evaluate whether SciKit Learn feature selection algorithms improve or worsen the accuracy and processing time of machine learning algorithms when used for network intrusion detection  ...  In our research paper, except for the ISCX-URL-2016 data-set, this trend follows.  ...  Our research forms the basis for future work on the development of a recommendation system for SK Learn machine learning and feature selection algorithms, in the application of intrusion detection.  ... 
dblp:conf/fair2/PowellBWA19 fatcat:vzb2s5d4ovcxvco4xazbk7bgm4
« Previous Showing results 1 — 15 out of 7,265 results