Wyniki wyszukiwania - BazTech

1

Towards explainability of hashtags in the light of Graph Spectral Clustering methods

Starosta Bartłomiej, Kłopotek Mieczysław A., Wierzchoń Sławomir T.

Studia Informatica : systems and information technology

|

2023

|

Vol. 2(29)

57--68

EN

Hashtags constitute an indispensable part of modern social media world. As more and more hashtags are invented, it becomes a necessity to create clusters of these hashtags. Nowadays, however, the clustering alone does not help the users. They are asking for justification or expressed in the modern AI language, the clustering has to be explainable. We discuss a novel approach to hashtag explanation via a measure of similarity between hashtags based on the Graph Spectral Analysis. The application of this similarity measure may go far beyond the classical clustering task. It can be used to provide with explanations for the hashtags. In this paper we propose such a novel view of the proposed hashtag similarity measure.

2

Application of explainable artificial intelligence in software bug classification

Chmielowski Łukasz, Kucharzak Michał, Burduk Robert

Informatyka, Automatyka, Pomiary w Gospodarce i Ochronie Środowiska

|

2023

|

T. 13, nr 1

14--17

EN

Fault management is an expensive process and analyzing data manually requires a lot of resources. Modern software bug tracking systems may be armed with automated bug report assignment functionality that facilitates bug classification or bug assignment to proper development group.For supporting decision systems, it would be beneficial to introduce information related to explainability. The purpose of this work is to evaluate the useof explainable artificial intelligence (XAI) in processes related to software development and bug classification based on bug reports created by either software testers or software users. The research was conducted on two different datasets. The first one is related to classification of security vs non-securitybug reports. It comes from a telecommunication company which develops software and hardware solutions for mobile operators. The second dataset contains a list of software bugs taken from an opensource project. In this dataset the task is to classify issues with one of following labels crash, memory, performance, and security. Studies on XAI-related algorithms show that there are no major differences in the results of the algorithms used when comparing them with others. Therefore, not only the users can obtain results with possible explanations or experts can verify model or its part before introducing into production, but also it does not provide degradation of accuracy. Studies showed that it could be put into practice, but it has not been done so far.

PL

Zarządzanie usterkami jest kosztownym procesem, a ręczna analiza danych wymaga znacznych zasobów. Nowoczesne systemy zarządzania usterkami w oprogramowaniu mogą być wyposażone w funkcję automatycznego przypisywania usterek, która ułatwia klasyfikację ustereklub przypisywanie usterek do właściwej grupy programistów. Dla wsparcia systemów decyzyjnych korzystne byłoby wprowadzenie informacji związanychz wytłumaczalnością. Celem tej pracy jest ocena możliwości wykorzystania wyjaśnialnej sztucznej inteligencji (XAI) w procesach związanych z tworzeniem oprogramowania i klasyfikacją usterek na podstawie raportów o usterkach tworzonych przez testerów oprogramowania lub użytkowników oprogramowania. Badania przeprowadzono na dwóch różnych zbiorach danych. Pierwszy z nich związany jest z klasyfikacją raportów o usterkach związanych z bezpieczeństwem i niezwiązanych z bezpieczeństwem. Dane te pochodzą od firmy telekomunikacyjnej, która opracowuje rozwiązania programowe i sprzętowe dla operatorów komórkowych. Drugi zestaw danych zawiera listę usterek oprogramowania pobranych z projektu opensource.W tym zestawie danych zadanie polega na sklasyfikowaniu problemów za pomocą jednej z następujących etykiet: awaria, pamięć, wydajnośći bezpieczeństwo. Badania przeprowadzone przy użyciu algorytmów związanych z XAI pokazują, że nie ma większych różnic w wynikach algorytmów stosowanych przy porównywaniu ich z innymi. Dzięki temu nie tylko użytkownicy mogą uzyskać wyniki z ewentualnymi wyjaśnieniami lub eksperci mogą zweryfikować model lub jego część przed wprowadzeniem do produkcji, ale także nie zapewnia to degradacji dokładności. Badania wykazały, że możnato zastosować w praktyce, ale do tej pory tego nie zrobiono.

3

Interpretability versus explainability: classification for understanding deep learning systems and models

Namatēvs Ivars, Sudars Kaspars, Dobrājs Artis

Computer Assisted Methods in Engineering and Science

|

2022

|

Vol. 29, no. 4

297--356

EN

The techniques of explainability and interpretability are not alternatives for many realworld problems, as recent studies often suggest. Interpretable machine learning is nota subset of explainable artificial intelligence or vice versa. While the former aims to build glass-box predictive models, the latter seeks to understand a black box using an explanatory model, a surrogate model, an attribution approach, relevance importance, or other statistics. There is concern that definitions, approaches, and methods do not match, leading to the inconsistent classification of deep learning systems and models for interpretation and explanation. In this paper, we attempt to systematically evaluate and classify the various basic methods of interpretability and explainability used in the field of deep learning.One goal of this paper is to provide specific definitions for interpretability and explainability in Deep Learning. Another goal is to spell out the various research methods for interpretability and explainability through the lens of the literature to create a systematic classifier for interpretability and explainability in deep learning. We present a classifier that summarizes the basic techniques and methods of explainability and interpretability models. The evaluation of the classifier provides insights into the challenges of developinga complete and unified deep learning framework for interpretability and explainability concepts, approaches, and techniques.

4

An adversarial explainable artificial intelligence (XAI) based approach for action forecasting

Dutta Vibekananda, Zielińska Teresa

Journal of Automation Mobile Robotics and Intelligent Systems

|

2020

|

Vol. 14, No. 4

3--10

EN

Despite the growing popularity of machine learning technology, vision‐based action recognition/forecasting systems are seen as black‐boxes by the user. The effecti‐ veness of such systems depends on the machine learning algorithms, it is difficult (or impossible) to explain the de‐ cisions making processes to the users. In this context, an approach that offers the user understanding of these re‐ asoning models is significant. To do this, we present an Explainable Artificial Intelligence (XAI) based approach to action forecasting using structured database and object affordances definition. The structured database is sup‐ porting the prediction process. The method allows to vi‐ sualize the components of the structured database. Later, the components of the base are used for forecasting the nominally possible motion goals. The object affordance explicated by the probability functions supports the se‐ lection of possible motion goals. The presented methodo‐ logy allows satisfactory explanations of the reasoning be‐ hind the inference mechanism. Experimental evaluation was conducted using the WUT‐18 dataset, the efficiency of the presented solution was compared to the other ba‐ seline algorithms.