Priscila Machado Vieira Lima

Zachary Susskind, Aman Arora, Igor Dantas Dos Santos Miranda et al.

Weightless Neural Networks (WNNs) are a class of machine learning model which use table lookups to perform inference. This is in contrast with Deep Neural Networks (DNNs), which use multiply-accumulate operations. State-of-the-art WNN architectures have a fraction of the implementation cost of DNNs, but still lag behind them on accuracy for common image recognition tasks. Additionally, many existing WNN architectures suffer from high memory requirements. In this paper, we propose a novel WNN architecture, BTHOWeN, with key algorithmic and architectural improvements over prior work, namely counting Bloom filters, hardware-friendly hashing, and Gaussian-based nonlinear thermometer encodings to improve model accuracy and reduce area and energy consumption. BTHOWeN targets the large and growing edge computing sector by providing superior latency and energy efficiency to comparable quantized DNNs. Compared to state-of-the-art WNNs across nine classification datasets, BTHOWeN on average reduces error by more than than 40% and model size by more than 50%. We then demonstrate the viability of the BTHOWeN architecture by presenting an FPGA-based accelerator, and compare its latency and resource usage against similarly accurate quantized DNN accelerators, including Multi-Layer Perceptron (MLP) and convolutional models. The proposed BTHOWeN models consume almost 80% less energy than the MLP models, with nearly 85% reduction in latency. In our quest for efficient ML on the edge, WNNs are clearly deserving of additional attention.

Rubens Lacerda Queiroz, Cabral Lima, Fabio Ferrentini Sampaio et al.

This study expands on previous work that introduced the AIcon2abs method (AI from Concrete to Abstract: Demystifying Artificial Intelligence to the general public), an innovative approach designed to increase public understanding of machine learning (ML) across diverse age groups, including K-12 students, and aims to evaluate its effectiveness. AIcon2Abs employs the WiSARD algorithm, a weightless neural network known for its simplicity, and user accessibility. WiSARD does not require Internet, making it ideal for non-technical users and resource-limited environments. This method enables participants to intuitively visualize and interact with ML processes through engaging, hands-on activities, as if they were the algorithms themselves. The method allows users to intuitively visualize and understand the internal processes of training and classification through practical activities. Once WiSARDs functionality does not require an Internet connection, it can learn effectively from a minimal dataset, even from a single example. This feature enables users to observe how the machine improves its accuracy incrementally as it receives more data. Moreover, WiSARD generates mental images representing what it has learned, highlighting essential features of the classified data. AIcon2abs was tested through a six-hour remote course with 34 Brazilian participants, including 5 children, 5 adolescents, and 24 adults. Data analysis was conducted from two perspectives: a mixed-method pre-experiment (including hypothesis testing), and a qualitative phenomenological analysis. Nearly all participants rated AIcon2abs positively, with the results demonstrating a high degree of satisfaction in achieving the intended outcomes. This research was approved by the CEP-HUCFF-UFRJ Research Ethics Committee.

Rubens Lacerda Queiroz, Fábio Ferrentini Sampaio, Cabral Lima et al.

Artificial Intelligence (AI) has been adopted in a wide range of domains. This shows the imperative need to develop means to endow common people with a minimum understanding of what AI means. Combining visual programming and WiSARD weightless artificial neural networks, this article presents a new methodology, AI from concrete to abstract (AIcon2abs), to enable general people (including children) to achieve this goal. The main strategy adopted by is to promote a demystification of artificial intelligence via practical activities related to the development of learning machines, as well as through the observation of their learning process. Thus, it is possible to provide subjects with skills that contributes to making them insightful actors in debates and decisions involving the adoption of artificial intelligence mechanisms. Currently, existing approaches to the teaching of basic AI concepts through programming treat machine intelligence as an external element/module. After being trained, that external module is coupled to the main application being developed by the learners. In the methodology herein presented, both training and classification tasks are blocks that compose the main program, just as the other programming constructs. As a beneficial side effect of AIcon2abs, the difference between a program capable of learning from data and a conventional computer program becomes more evident. In addition, the simplicity of the WiSARD weightless artificial neural network model enables easy visualization and understanding of training and classification tasks internal realization.

Tarek R. Besold, Artur d'Avila Garcez, Sebastian Bader et al.

The study and understanding of human behaviour is relevant to computer science, artificial intelligence, neural computation, cognitive science, philosophy, psychology, and several other areas. Presupposing cognition as basis of behaviour, among the most prominent tools in the modelling of behaviour are computational-logic systems, connectionist models of cognition, and models of uncertainty. Recent studies in cognitive science, artificial intelligence, and psychology have produced a number of cognitive models of reasoning, learning, and language that are underpinned by computation. In addition, efforts in computer science research have led to the development of cognitive computational systems integrating machine learning and automated reasoning. Such systems have shown promise in a range of applications, including computational biology, fault diagnosis, training and assessment in simulators, and software verification. This joint survey reviews the personal ideas and views of several researchers on neural-symbolic learning and reasoning. The article is organised in three parts: Firstly, we frame the scope and goals of neural-symbolic computation and have a look at the theoretical foundations. We then proceed to describe the realisations of neural-symbolic computation, systems, and applications. Finally we present the challenges facing the area and avenues for further research.