Search results for: explainable AI

Authors: Mufutau Ayinla Abdul-Yakeen

Abstract:

An analysis of the five most used numerals and a proposal for the adoption of a Universally Acceptable Numerals (UAN), came up as a result of the researchers inquisitiveses of the need for a set of numerals that is universally accepted. The researcher sought for the meaning of the first letter, “Nun”, “ن”, of the first verse of Suratul-Kalam (Chapter of the Pen), the Sixty-Eighth Chapter of the Holy Qur'an. It was observed that there was no universally accepted, economical, explainable, linkable and consistent set of numerals used by all scientists up till the moment of making this enquiry. As a theoretical paper, explanatory method is used to review five of the most used numerals (Tally Marks, Roman Figure, Hindu-Arabic, Arabic, and Chinese) and the urgent need for a universally accepted, economical, explainable, linkable and consistent set of numerals arises. The study discovers: ^., I, \, _, L, U, =, C, O, 9, and 1.; to be used as numeral 0, 1, 2, 3, 4, 5, 6, 7, 8, 9 and 10 respectively; as a set of universally acceptable, economical, explainable, linkable, sustainable, convertible and consistent set of numerals that originates from Islam. They can be called Islameconumerals or UAN. With UAN, everything dropped, written, drawn and/or scribbled has meaning(s) as postulated by the first verse of Qur'an 68 and everyone can easily document all figures within the shortest period. It is suggested that there should be a discipline called Numeralnomics (Study of optimum utilization of Numerals) and everybody should start using the UAN, now, in order in know their strengths and weaknesses so as to suggest a better and acceptable set of numerals for the interested readers. Similarly study can be conducted for the alphabets.

Keywords: acceptable, economical, explainable, Islameconumerals, numeralnomics

Procedia PDF Downloads 296

Authors: Jacquelyn Burkell, Jane Bailey

Abstract:

Artificial intelligence (AI) solutions deployed within the justice system face the critical task of providing acceptable explanations for decisions or actions. These explanations must satisfy the joint criteria of public and professional accountability, taking into account the perspectives and requirements of multiple stakeholders, including judges, lawyers, parties, witnesses, and the general public. This research project analyzes and integrates two existing literature on explanations in order to propose guidelines for explainable AI in the justice system. Specifically, we review three bodies of literature: (i) explanations of the purpose and function of 'explainable AI'; (ii) the relevant case law, judicial commentary and legal literature focused on the form and function of reasons for judicial decisions; and (iii) the literature focused on the psychological and sociological functions of these reasons for judicial decisions from the perspective of the public. Our research suggests that while judicial ‘reasons’ (arguably accurate descriptions of the decision-making process and factors) do serve similar explanatory functions as those identified in the literature on 'explainable AI', they also serve an important ‘justification’ function (post hoc constructions that justify the decision that was reached). Further, members of the public are also looking for both justification and explanation in reasons for judicial decisions, and that the absence of either feature is likely to contribute to diminished public confidence in the legal system. Therefore, artificially automated judicial decision-making systems that simply attempt to document the process of decision-making are unlikely in many cases to be useful to and accepted within the justice system. Instead, these systems should focus on the post-hoc articulation of principles and precedents that support the decision or action, especially in cases where legal subjects’ fundamental rights and liberties are at stake.

Keywords: explainable AI, judicial reasons, public accountability, explanation, justification

Procedia PDF Downloads 97

Authors: Esther Chiramal, Kelvin Soh Boon Kai

Abstract:

Explainable AI is a strong strategy implemented to understand complex black-box model predictions in a human-interpretable language. It provides the evidence required to execute the use of trustworthy and reliable AI systems. On the other hand, however, it also opens the door to locating possible vulnerabilities in an AI model. Traditional adversarial text attack uses word substitution, data augmentation techniques, and gradient-based attacks on powerful pre-trained Bidirectional Encoder Representations from Transformers (BERT) variants to generate adversarial sentences. These attacks are generally white-box in nature and not practical as they can be easily detected by humans e.g., Changing the word from “Poor” to “Rich”. We proposed a simple yet effective Grey-box cum Black-box approach that does not require the knowledge of the model while using a set of surrogate Transformer/BERT models to perform the attack using Explainable AI techniques. As Transformers are the current state-of-the-art models for almost all Natural Language Processing (NLP) tasks, an attack generated from BERT1 is transferable to BERT2. This transferability is made possible due to the attention mechanism in the transformer that allows the model to capture long-range dependencies in a sequence. Using the power of BERT generalisation via attention, we attempt to exploit how transformers learn by attacking a few surrogate transformer variants which are all based on a different architecture. We demonstrate that this approach is highly effective to generate semantically good sentences by changing as little as one word that is not detectable by humans while still fooling other BERT models.

Keywords: BERT, explainable AI, Grey-box text attack, transformer

Procedia PDF Downloads 112

Authors: David Pham, Yongfeng Zhang

Abstract:

Graphs are an important structure for data storage and computation. Recent years have seen the success of deep learning on graphs such as Graph Neural Networks (GNN) on various data mining and machine learning tasks. However, most of the deep learning models on graphs cannot easily explain their predictions and are thus often labelled as “black boxes.” For example, Graph Attention Network (GAT) is a frequently used GNN architecture, which adopts an attention mechanism to carefully select the neighborhood nodes for message passing and aggregation. However, it is difficult to explain why certain neighbors are selected while others are not and how the selected neighbors contribute to the final classification result. In this paper, we present a graph learning model called Explainable Graph Attention Network (XGAT), which integrates graph attention modeling and explainability. We use a single model to target both the accuracy and explainability of problem spaces and show that in the context of graph attention modeling, we can design a unified neighborhood selection strategy that selects appropriate neighbor nodes for both better accuracy and enhanced explainability. To justify this, we conduct extensive experiments to better understand the behavior of our model under different conditions and show an increase in both accuracy and explainability.

Keywords: explainable AI, graph attention network, graph neural network, node classification

Procedia PDF Downloads 140

Authors: Antti Nurminen, Avleen Malhi

Abstract:

Significant progress in intelligent systems in combination with exceedingly wide application domains having machine learning as the core technology are usually opaque, non-intuitive, and commonly complex for human users. We use innovative IoT technology which monitors and analyzes moisture, humidity, luminosity and temperature levels to assist end users for optimization of environmental conditions for their houseplants. For plant health monitoring, we construct a system yielding the Normalized Difference Vegetation Index (NDVI), supported by visual validation by users. We run the system for a selected plant, basil, in varying environmental conditions to cater for typical home conditions, and bootstrap our AI with the acquired data. For end users, we implement a web based user interface which provides both instructions and explanations.

Keywords: explainable artificial intelligence, intelligent agent, IoT, NDVI

Procedia PDF Downloads 134

Authors: Nicolas M. Beleski, Gustavo A. G. Lugo

Abstract:

Robotic agents are taking more and increasingly important roles in society. In order to make these robots and agents more autonomous and efficient, their systems have grown to be considerably complex and convoluted. This growth in complexity has led recent researchers to investigate forms to explain the AI behavior behind these systems in search for more trustworthy interactions. A current problem in explainable AI is the inner workings with the logic inference process and how to conduct a sensibility analysis of the process of valuation and alteration of beliefs. In a social HRI (human-robot interaction) setup, theory of mind is crucial to ease the intentionality gap and to achieve that we should be able to infer over observed human behaviors, such as cases of cognitive dissonance. One specific case inspired in human cognition is the role emotions play on our belief system and the effects caused when observed behavior does not match the expected outcome. In such scenarios emotions can make a person wrongly assume the antecedent P for an observed consequent Q, and as a result, incorrectly assert that P is true. This form of cognitive dissonance where an unproven cause is taken as truth induces changes in the belief base which can directly affect future decisions and actions. If we aim to be inspired by human thoughts in order to apply levels of theory of mind to these artificial agents, we must find the conditions to replicate these observable cognitive mechanisms. To achieve this, a computational architecture is proposed to model the modulation effect emotions have on the belief system and how it affects logic inference process and consequently the decision making of an agent. To validate the model, an experiment based on the prisoner's dilemma is currently under development. The hypothesis to be tested involves two main points: how emotions, modeled as internal argument strength modulators, can alter inference outcomes, and how can explainable outcomes be produced under specific forms of cognitive dissonance.

Keywords: cognitive architecture, cognitive dissonance, explainable ai, sensitivity analysis, theory of mind

Procedia PDF Downloads 111

Authors: Raphael Tuor, Denis Lalanne

Abstract:

The ATM research community is missing suitable tools to design, test, and validate new UI prototypes. Important stakes underline the implementation of both DSS and XAI methods into current systems. ML-based DSS are gaining in relevance as ATFM becomes increasingly complex. However, these systems only prove useful if a human can understand them, and thus new XAI methods are needed. The human-machine dyad should work as a team and should understand each other. We present xSky, a configurable benchmark tool that allows us to compare different versions of an ATC interface in conflict detection tasks. Our main contributions to the ATC research community are (1) a conflict detection task simulator (xSky) that allows to test the applicability of visual prototypes on scenarios of varying difficulty and outputting relevant operational metrics (2) a theoretical approach to the explanations of AI-driven trajectory predictions. xSky addresses several issues that were identified within available research tools. Researchers can configure the dimensions affecting scenario difficulty with a simple CSV file. Both the content and appearance of the XAI elements can be customized in a few steps. As a proof-of-concept, we implemented an XAI prototype inspired by the maritime field.

Keywords: air traffic control, air traffic simulation, conflict detection, explainable artificial intelligence, explainability, human-automation collaboration, human factors, information visualization, interpretability, trajectory prediction

Procedia PDF Downloads 131

Authors: Jixiang Lei, Alexander Fuchs, Franz Pernkopf, Katrin Ellermann

Abstract:

Within the Space@Sea project, funded by the Horizon 2020 program, an island consisting of multiple platforms was designed. The platforms are connected by ropes and fenders. The connection is critical with respect to the safety of the whole system. Therefore, fault detection systems are investigated, which could detect early warning signs for a possible failure in the connection elements. Previously, a model-based method called Extended Kalman Filter was developed to detect the reduction of rope stiffness. This method detected several types of faults reliably, but some types of faults were much more difficult to detect. Furthermore, the model-based method is sensitive to environmental noise. When the wave height is low, a long time is needed to detect a fault and the accuracy is not always satisfactory. In this sense, it is necessary to develop a more accurate and robust technique that can detect all rope faults under a wide range of operational conditions. Inspired by this work on the Space at Sea design, we introduce a fault diagnosis method based on deep neural networks. Our method cannot only detect rope degradation by using the acceleration data from each platform but also estimate the contributions of the specific acceleration sensors using methods from explainable AI. In order to adapt to different operational conditions, the domain adaptation technique DANN is applied. The proposed model can accurately estimate rope degradation under a wide range of environmental conditions and help users understand the relationship between the output and the contributions of each acceleration sensor.

Keywords: fault diagnosis, deep learning, domain adaptation, explainable AI

Procedia PDF Downloads 153

Authors: Mpho Mokoatle, Darlington Mapiye, James Mashiyane, Stephanie Muller, Gciniwe Dlamini

Abstract:

Great advances in high-throughput sequencing technologies have resulted in availability of huge amounts of sequencing data in public and private repositories, enabling a holistic understanding of complex biological phenomena. Sequence data are used for a wide range of applications such as gene annotations, expression studies, personalized treatment and precision medicine. However, this rapid growth in sequence data poses a great challenge which calls for novel data processing and analytic methods, as well as huge computing resources. In this work, a machine and statistical learning approach for DNA sequence classification based on $k$-mer representation of sequence data is proposed. The approach is tested using whole genome sequences of Mycobacterium tuberculosis (MTB) isolates to (i) reduce the size of genomic sequence data, (ii) identify an optimum size of k-mers and utilize it to build classification models, (iii) predict the phenotype from whole genome sequence data of a given bacterial isolate, and (iv) demonstrate computing challenges associated with the analysis of whole genome sequence data in producing interpretable and explainable insights. The classification models were trained on 104 whole genome sequences of MTB isoloates. Cluster analysis showed that k-mers maybe used to discriminate phenotypes and the discrimination becomes more concise as the size of k-mers increase. The best performing classification model had a k-mer size of 10 (longest k-mer) an accuracy, recall, precision, specificity, and Matthews Correlation coeffient of 72.0%, 80.5%, 80.5%, 63.6%, and 0.4 respectively. This study provides a comprehensive approach for resampling whole genome sequencing data, objectively selecting a k-mer size, and performing classification for phenotype prediction. The analysis also highlights the importance of increasing the k-mer size to produce more biological explainable results, which brings to the fore the interplay that exists amongst accuracy, computing resources and explainability of classification results. However, the analysis provides a new way to elucidate genetic information from genomic data, and identify phenotype relationships which are important especially in explaining complex biological mechanisms.

Keywords: AWD-LSTM, bootstrapping, k-mers, next generation sequencing

Procedia PDF Downloads 133

Authors: Darlington Mapiye, Mpho Mokoatle, James Mashiyane, Stephanie Muller, Gciniwe Dlamini

Abstract:

Great advances in high-throughput sequencing technologies have resulted in availability of huge amounts of sequencing data in public and private repositories, enabling a holistic understanding of complex biological phenomena. Sequence data are used for a wide range of applications such as gene annotations, expression studies, personalized treatment and precision medicine. However, this rapid growth in sequence data poses a great challenge which calls for novel data processing and analytic methods, as well as huge computing resources. In this work, a machine and statistical learning approach for DNA sequence classification based on k-mer representation of sequence data is proposed. The approach is tested using whole genome sequences of Mycobacterium tuberculosis (MTB) isolates to (i) reduce the size of genomic sequence data, (ii) identify an optimum size of k-mers and utilize it to build classification models, (iii) predict the phenotype from whole genome sequence data of a given bacterial isolate, and (iv) demonstrate computing challenges associated with the analysis of whole genome sequence data in producing interpretable and explainable insights. The classification models were trained on 104 whole genome sequences of MTB isoloates. Cluster analysis showed that k-mers maybe used to discriminate phenotypes and the discrimination becomes more concise as the size of k-mers increase. The best performing classification model had a k-mer size of 10 (longest k-mer) an accuracy, recall, precision, specificity, and Matthews Correlation coeffient of 72.0 %, 80.5 %, 80.5 %, 63.6 %, and 0.4 respectively. This study provides a comprehensive approach for resampling whole genome sequencing data, objectively selecting a k-mer size, and performing classification for phenotype prediction. The analysis also highlights the importance of increasing the k-mer size to produce more biological explainable results, which brings to the fore the interplay that exists amongst accuracy, computing resources and explainability of classification results. However, the analysis provides a new way to elucidate genetic information from genomic data, and identify phenotype relationships which are important especially in explaining complex biological mechanisms

Keywords: AWD-LSTM, bootstrapping, k-mers, next generation sequencing

Procedia PDF Downloads 123

Authors: Hassan Eshkiki, Benjamin Mora

Abstract:

The Neighbor Migrating Generator is a simple and efficient approach to finding the closest potential neighbor(s) with a different label for a given instance and so without the need to calibrate any kernel settings at all. This allows determining and explaining the most important features that will influence an AI model. It can be used to either migrate a specific sample to the class decision boundary of the original model within a close neighborhood of that sample or identify global features that can help localising neighbor classes. The proposed technique works by minimizing a loss function that is divided into two components which are independently weighted according to three parameters α, β, and ω, α being self-adjusting. Results show that this approach is superior to past techniques when detecting the smallest changes in the feature space and may also point out issues in models like over-fitting.

Keywords: explainable AI, EX AI, feature importance, counterfactual explanations

Procedia PDF Downloads 121

Authors: Haonan Hu, Shuge Lei, Dasheng Sun, Huabin Zhang, Kehong Yuan, Jian Dai, Jijun Tang

Abstract:

This paper focuses on the classification task of breast ultrasound images and conducts research on the reliability measurement of classification results. A dual-channel evaluation framework was developed based on the proposed inference reliability and predictive reliability scores. For the inference reliability evaluation, human-aligned and doctor-agreed inference rationals based on the improved feature attribution algorithm SP-RISA are gracefully applied. Uncertainty quantification is used to evaluate the predictive reliability via the test time enhancement. The effectiveness of this reliability evaluation framework has been verified on the breast ultrasound clinical dataset YBUS, and its robustness is verified on the public dataset BUSI. The expected calibration errors on both datasets are significantly lower than traditional evaluation methods, which proves the effectiveness of the proposed reliability measurement.

Keywords: medical imaging, ultrasound imaging, XAI, uncertainty measurement, trustworthy AI

Procedia PDF Downloads 58

Authors: Yong Wee Foo, Lai Meng Tang

Abstract:

Artificial Intelligence (AI), particularly the usage of deep neural networks for hierarchical representations from data, has found numerous complex applications across various domains, including computer vision, robotics, autonomous vehicles, and other scientific fields. However, their inherent “black box” nature can sometimes make it challenging for early researchers or school students of various levels to comprehend and trust the results they produce. Consequently, there has been a growing demand for reliable visualization tools in engineering and science education to help learners understand, trust, and explain a deep learning network. This has led to a notable emphasis on the visualization of AI in the research community in recent years. AI visualization tools are increasingly being adopted to significantly improve the comprehension of complex topics in deep learning. This paper proposes a novel approach to empower students to actively explore the inner workings of deep neural networks by combining the student-centered learning approach of flipped classroom models with the investigative power of AI visualization tools, namely, the TensorFlow Playground, the Local Interpretable Model-agnostic Explanations (LIME), and the SHapley Additive exPlanations (SHAP), for delivering an AI education curriculum. Combining the two factors is vital in fostering ownership, responsibility, and critical thinking skills in the age of AI.

Keywords: deep learning, explainable AI, AI visualization, representation learning

Procedia PDF Downloads 35

Authors: Funmilayo Abiola Opadoja, Samuel Olukayode Awotona

Abstract:

Nursing services costing has been a major interest to nurses for a long period of time. Determination of nursing costing is germane in order to show the effectiveness of nursing practice in an improved and affordable health care delivery system. This has been a major concern of managers that have the mind of quality and affordable health services. The treatment or intervention should be considered as ‘product’ of nursing care and should provide an explainable term for billing. The study was non-experimental, descriptive and went about eliciting the views of nurses on costing nursing care at two public hospitals namely: University College Hospital and Adeoyo Maternity Teaching Hospital. The questionnaire was the instrument used in eliciting nurse’s response. It was administered randomly on 300 selected respondents across various wards within the hospitals. The data was collected and analysed using SPSS20.0 to generate frequency, and cross-tabulations to explore the statistical relationship between variables. The result shows that 89.2% of the respondents viewed costing of nursing care as an important issued to be looked into. The study concluded that nursing care costing is germane to enhancing the status and imagery of the nurses, it is essential because it would enhance the performance of nurses in discharging their duties. There is need to have a procedural manual agreed on by nursing practitioner on costing of each care given.

Keywords: costing, health care delivery system, intervention, nursing care, practitioner

Procedia PDF Downloads 282

Authors: Ali Bahri Lubis

Abstract:

Tryptophan oxidation by Heme-dioxygenase enzyme is the initial rate-limiting step in the kynurenine pathway, which leads to the formation of NADH and dangerous metabolites, implicating several severe diseases such as Parkinson’s Disease, Huntington's Disease, poliomyelitis and cataract. This oxidation, generally, allows tryptophan to convert to N-Formylkynurenine (NFK). Observing the catalytic mechanism of Heme dioxygenase in tryptophan oxidation has been a debatably scientific interest since no one has yet proven the mechanism obviously. In this research we have attempted to prove mechanistic steps of tryptophan oxidation via human indoleamine dioxygenase (h-IDO) utilising various substrates: L-tryptophan, L-tryptophan (indole-ring-2-¹³C), L-fully-labelled¹³C-tryptophan, L-N-methyl-tryptophan, L-tryptophanol and 2-amino-3-(benzo(b)thiophene-3-yl) propanoic acid. All enzyme assay experiments were measured using a UV-Vis spectrophotometer, LC-MS, 1H-NMR and HSQC. We also successfully synthesised enzyme products as our control in NMR measurements. The result exhibited that all substrates produced N-formyl kynurenine (NFK), and a side, the minor product of hydroxypyrrolloindoleamine carboxylic acid (HPIC) in cis and trans isomer, except 1-methyl tryptophan only generating cis HPIC. Interestingly, L- tryptophanol was oxidised to form HPIC derivative as a major product and 5-hydroxy tryptophan was converted to NFK derivative instead without any HPIC derivative. The bizarre result of oxidation underwent in 2-amino-3-(benzo(b)thiophene-3-yl) propanoic acid, which produced epoxide cyclic. Those phenomena have been explainable in our research based on the proposed mechanism of how tryptophan is oxidised by human indoleamine dioxygenase.

Keywords: tryptophan oxidation, heme-dioxygenases, human indoleamine dioxygenases, N-formylkynurenine, hydroxypyrroloindoleamine carboxylic acid

Procedia PDF Downloads 38

Authors: Alban Igwe, Ijeoma Kalu, Alloy Ezirim

Abstract:

Food insecurity is a global problem, and the search for food security has assumed a central stage in the global development agenda as the United Nations currently placed zero hunger as a goal number in its sustainable development goals. Nigeria currently ranks 107th out of 113 countries in the global food security index (GFSI), a metric that defines a country's ability to furnish its citizens with food and nutrients for healthy living. Paradoxically, Nigeria is a global leader in food production, ranking 1st in yam (over 70% of global output), beans (over 41% of global output), cassava (20% of global output) and shea nuts, where it commands 53% of global output. Furthermore, it ranks 2nd in millet, sweet potatoes, and cashew nuts. It is Africa's largest producer of rice. So, it is apparent that Nigeria's food insecurity woes must relate to a factor other than food production. We investigated the nexus between food security and downstream supply chain losses in the yam industry with secondary data from the Food and Agricultural Organization (FAOSTAT) and the National Bureau of Statics for the decade 2012-2021. In analyzing the data, multiple regression techniques were used, and findings reveal that downstream losses have a strong positive correlation with food security (r = .763*) and a 58.3% variation in food security is explainable by post-downstream supply chain food losses. The study discovered that yam supply chain losses within the period under review averaged 50.6%, suggestive of the fact that downstream supply chain losses are the drainpipe and the major source of food insecurity in Nigeria. Therefore, the study concluded that there is a significant relationship between downstream supply chain losses and food insecurity and recommended the establishment of food supply chain structures and policies to enhance food security in Nigeria.

Keywords: food security, downstream supply chain losses, yam, nigeria, supply chain

Procedia PDF Downloads 60

Authors: Satendra Pal Singh, Dalip Kr. Kakru, Jyoti Mishra, Rajesh Thakur, Tarana Sarwat

Abstract:

COVID-19 is still an intrigue as far as morbidity or mortality is concerned. The rate of recovery varies from person to person, & it depends upon the accessibility of the healthcare system and the roles played by the physicians and caregivers. It is envisaged that with the passage of time, people would become immune to this virus, and those who are vulnerable would sustain themselves with the help of vaccines. The proposed study deals with the severeness of COVID-19 is associated with some specific biomarkers linked to correlate age and gender. We will be assessing the overall homeostasis of the persons who were affected by the coronavirus infection and also of those who recovered from it. Some people show more severe effects, while others show very mild symptoms, however, they show low CT values. Thus far, it is unclear why the new strain of Covid has different effects on different people in terms of age, gender, and ABO blood typing. According to data, the fatality rate with heart disease was 10.5 percent, 7.3 percent were diabetic, and 6 percent who are already infected from other comorbidities. However, some COVID-19 cases are worse than others & it is not fully explainable as of date. Overall data show that the ABO blood group is effective or prone to the risk of SARS-COV2 infection, while another study also shows the phenotypic effects of the blood group related to covid. It is an accepted fact that females have more strong immune systems than males, which may be related to the fact that females have two ‘X’ chromosomes, which might contain a more effective immunity booster gene on the X chromosome, and are capable to protect the female. Also specific sex hormones also induce a better immune response in a specific gender. This calls for in-depth analysis to be able to gain insight into this dilemma. COVID-19 is still not fully characterized, and thus we are not very familiar with its biology, mode of infection, susceptibility, and overall viral load in the human body. How many virus particles are needed to infect a person? How, then, comorbidity contribute to coronavirus infection? Since the emergence of this virus in 2020, a large number of papers have been published, and seemingly, vaccines have been prepared. But still, a large number of questions remain unanswered. The proneness of humans for infection by covid-19 needs to be established to be able to develop a better strategy to fight this virus. Our study will be on the Impact of demography on the Severity of covid-19 infection & at the same time, will look into gender-specific sensitivity of Covid-19 and the Operational variation of different biochemical markers in Covid-19 positive patients. Besides, we will be studying the co-relation, if any, of COVID severity & ABO Blood group type and the occurrence of the most common blood group type amongst positive patience.

Keywords: coronavirus, ABO blood group, age, gender

Procedia PDF Downloads 69

Authors: Arkadiusz Gut, Robert Mirski

Abstract:

The nativist-constructivist debate constitutes a considerable part of current research on mindreading. Peter Carruthers and his colleagues are known for their nativist position in the debate and take issue with constructivist views proposed by other researchers, with Henry Wellman, Alison Gopnik, and Ian Apperly at the forefront. More specifically, Carruthers together with Evan Westra propose a nativistic explanation of Theory of Mind Scale study results that Wellman et al. see as supporting constructivism. While allowing for development of the innate mindreading system, Westra and Carruthers base their argumentation essentially on a competence-performance gap, claiming that cross-cultural differences in Theory of Mind Scale progression as well as discrepancies between infants’ and toddlers’ results on verbal and non-verbal false-belief tasks are fully explainable in terms of acquisition of other, pragmatic, cognitive developments, which are said to allow for an expression of the innately present Theory of Mind understanding. The goal of the present paper is to bring together arguments against the view offered by Westra and Carruthers. It will be shown that even though Carruthers et al.’s interpretation has not been directly controlled for in Wellman et al.’s experiments, there are serious reasons to dismiss such nativistic views which Carruthers et al. advance. The present paper discusses the following issues that undermine Carruthers et al.’s nativistic conception: (1) The concept of innateness is argued to be developmentally inaccurate; it has been dropped in many biological sciences altogether and many developmental psychologists advocate for doing the same in cognitive psychology. Reality of development is a complex interaction of changing elements that is belied by the simplistic notion of ‘the innate.’ (2) The purported innate mindreading conceptual system posited by Carruthers ascribes adult-like understanding to infants, ignoring the difference between first- and second-order understanding, between what can be called ‘presentation’ and ‘representation.’ (3) Advances in neurobiology speak strongly against any inborn conceptual knowledge; neocortex, where conceptual knowledge finds its correlates, is said to be largely equipotential at birth. (4) Carruthers et al.’s interpretations are excessively charitable; they extend results of studies done with 15-month-olds to conclusions about innateness, whereas in reality at that age there has been plenty of time for construction of the skill. (5) Looking-time experiment paradigm used in non-verbal false belief tasks that provide the main support for Carruthers’ argumentation has been criticized on methodological grounds. In the light of the presented arguments, nativism in theory of mind research is concluded to be an untenable position.

Keywords: development, false belief, mindreading, nativism, theory of mind

Procedia PDF Downloads 185

Authors: Henning Lahmann, Benjamyn I. Scott, Bart Custers

Abstract:

Suboptimal adoption of AI in defence organisations carries risks for the protection of the freedom, safety, and security of society. Despite the vast opportunities that defence AI-technology presents, there are also a variety of ethical, legal, and societal concerns. To ensure the successful use of AI technology by the military, ethical, legal, and societal aspects (ELSA) need to be considered, and their concerns continuously addressed at all levels. This includes ELSA considerations during the design, manufacturing and maintenance of AI-based systems, as well as its utilisation via appropriate military doctrine and training. This raises the question how defence organisations can remain strategically competitive and at the edge of military innovation, while respecting the values of its citizens. This paper will explain the set-up and share preliminary results of a 4-year research project commissioned by the National Research Council in the Netherlands on the ethical, legal, and societal aspects of AI in defence. The project plans to develop a future-proof, independent, and consultative ecosystem for the responsible use of AI in the defence domain. In order to achieve this, the lab shall devise a context-dependent methodology that focuses on the ‘analysis’, ‘design’ and ‘evaluation’ of ELSA of AI-based applications within the military context, which include inter alia unmanned aircraft. This is bolstered as the Lab also recognises and complements the existing methods in regards to human-machine teaming, explainable algorithms, and value-sensitive design. Such methods will be modified for the military context and applied to pertinent case-studies. These case-studies include, among others, the application of autonomous robots (incl. semi- autonomous) and AI-based methods against cognitive warfare. As the perception of the application of AI in the military context, by both society and defence personnel, is important, the Lab will study how these perceptions evolve and vary in different contexts. Furthermore, the Lab will monitor – as they may influence people’s perception – developments in the global technological, military and societal spheres. Although the emphasis of the research project is on different forms of AI in defence, it focuses on several case studies. One of these case studies is on unmanned aircraft, which will also be the focus of the paper. Hence, ethical, legal, and societal aspects of unmanned aircraft in the defence domain will be discussed in detail, including but not limited to privacy issues. Typical other issues concern security (for people, objects, data or other aircraft), privacy (sensitive data, hindrance, annoyance, data collection, function creep), chilling effects, PlayStation mentality, and PTSD.

Keywords: autonomous weapon systems, unmanned aircraft, human-machine teaming, meaningful human control, value-sensitive design

Procedia PDF Downloads 71

Authors: Ola Hall, Ibrahim Wahab, Thorsteinn Rognvaldsson, Mattias Ohlsson

Abstract:

The subfield of poverty and welfare estimation that applies machine learning tools and methods on satellite imagery is a nascent but rapidly growing one. This is in part driven by the sustainable development goal, whose overarching principle is that no region is left behind. Among other things, this requires that welfare levels can be accurately and rapidly estimated at different spatial scales and resolutions. Conventional tools of household surveys and interviews do not suffice in this regard. While they are useful for gaining a longitudinal understanding of the welfare levels of populations, they do not offer adequate spatial coverage for the accuracy that is needed, nor are their implementation sufficiently swift to gain an accurate insight into people and places. It is this void that satellite imagery fills. Previously, this was near-impossible to implement due to the sheer volume of data that needed processing. Recent advances in machine learning, especially the deep learning subtype, such as deep neural networks, have made this a rapidly growing area of scholarship. Despite their unprecedented levels of performance, such models lack transparency and explainability and thus have seen limited downstream applications as humans generally are apprehensive of techniques that are not inherently interpretable and trustworthy. While several studies have demonstrated the superhuman performance of AI models, none has directly compared the performance of such models and human readers in the domain of poverty studies. In the present study, we directly compare the performance of human readers and a DL model using different resolutions of satellite imagery to estimate the welfare levels of demographic and health survey clusters in Tanzania, using the wealth quintile ratings from the same survey as the ground truth data. The cluster-level imagery covers all 608 cluster locations, of which 428 were classified as rural. The imagery for the human readers was sourced from the Google Maps Platform at an ultra-high resolution of 0.6m per pixel at zoom level 18, while that of the machine learning model was sourced from the comparatively lower resolution Sentinel-2 10m per pixel data for the same cluster locations. Rank correlation coefficients of between 0.31 and 0.32 achieved by the human readers were much lower when compared to those attained by the machine learning model – 0.69-0.79. This superhuman performance by the model is even more significant given that it was trained on the relatively lower 10-meter resolution satellite data while the human readers estimated welfare levels from the higher 0.6m spatial resolution data from which key markers of poverty and slums – roofing and road quality – are discernible. It is important to note, however, that the human readers did not receive any training before ratings, and had this been done, their performance might have improved. The stellar performance of the model also comes with the inevitable shortfall relating to limited transparency and explainability. The findings have significant implications for attaining the objective of the current frontier of deep learning models in this domain of scholarship – eXplainable Artificial Intelligence through a collaborative rather than a comparative framework.

Keywords: poverty prediction, satellite imagery, human readers, machine learning, Tanzania

Procedia PDF Downloads 70

Authors: Ursula Das, Manavjit Singh Dhindsa, Kshirasagar Naik, Marzia Zaman, Richard Purcell, Srinivas Sampalli, Abdul Mutakabbir, Chung-Horng Lung, Thambirajah Ravichandran

Abstract:

Wildland fires, also known as forest fires or wildfires, are exhibiting an alarming surge in frequency in recent times, further adding to its perennial global concern. Forest fires often lead to devastating consequences ranging from loss of healthy forest foliage and wildlife to substantial economic losses and the tragic loss of human lives. Despite the existence of substantial literature on the detection of active forest fires, numerous potential research avenues in forest fire management, such as preventative measures and ancillary effects of forest fires, remain largely underexplored. This paper undertakes a systematic review of these underexplored areas in forest fire research, meticulously categorizing them into distinct phases, namely pre-fire, during-fire, and post-fire stages. The pre-fire phase encompasses the assessment of fire risk, analysis of fuel properties, and other activities aimed at preventing or reducing the risk of forest fires. The during-fire phase includes activities aimed at reducing the impact of active forest fires, such as the detection and localization of active fires, optimization of wildfire suppression methods, and prediction of the behavior of active fires. The post-fire phase involves analyzing the impact of forest fires on various aspects, such as the extent of damage in forest areas, post-fire regeneration of forests, impact on wildlife, economic losses, and health impacts from byproducts produced during burning. A comprehensive understanding of the three stages is imperative for effective forest fire management and mitigation of the impact of forest fires on both ecological systems and human well-being. Artificial intelligence and machine learning (AI/ML) methods have garnered much attention in the cyber-physical systems domain in recent times leading to their adoption in decision-making in diverse applications including disaster management. This paper explores the current state of AI/ML applications for managing the activities in the aforementioned phases of forest fire. While conventional machine learning and deep learning methods have been extensively explored for the prevention, detection, and management of forest fires, a systematic classification of these methods into distinct AI research domains is conspicuously absent. This paper gives a comprehensive overview of the state of forest fire research across more recent and prominent AI/ML disciplines, including big data, classical machine learning, computer vision, explainable AI, generative AI, natural language processing, optimization algorithms, and time series forecasting. By providing a detailed overview of the potential areas of research and identifying the diverse ways AI/ML can be employed in forest fire research, this paper aims to serve as a roadmap for future investigations in this domain.

Keywords: artificial intelligence, computer vision, deep learning, during-fire activities, forest fire management, machine learning, pre-fire activities, post-fire activities

Procedia PDF Downloads 37