Researchers led by Concordia University have developed a blockchain-based framework that aims to democratize complex AI tasks, making advanced deep reinforcement learning more accessible and transparent for small businesses and individuals through crowdsourcing and smart contracts.
Tomorrow’s work environment will rely heavily on vast amounts of data, creating an unprecedented need for sophisticated AI systems, well-trained AI personnel and efficient servers. While tech giants have the means to harness this technology, small and medium-sized enterprises (SMEs) often struggle to access such resources.
In a bid to level the playing field, a team of researchers led by Concordia University has developed a framework that simplifies access to deep reinforcement learning (DRL), a subset of machine learning combining deep learning and reinforcement learning. Published in the journal Information Sciences, this framework is designed to assist industries as varied as gaming, robotics, health care and finance.
“Crowdsourcing the process of training and designing DRL makes the process more transparent and more accessible,” lead author Ahmed Alagha, a doctoral candidate at Concordia’s Gina Cody School of Engineering and Computer Science, said in a news release. “With this framework, anyone can sign up and build a history and profile. Based on their expertise, training and ratings, they can be allocated tasks that users are requesting.”
The framework pairs developers, companies and individuals with specific AI needs with service providers who possess the required resources, expertise and models. By leveraging blockchain technology and smart contracts, it ensures transparent and secure transactions.
Jamal Bentahar, a professor at the Concordia Institute for Information Systems Engineering and co-author of the study, emphasized the broader impact.
“To train a DRL model, you need computational resources that are not available to everyone. You also need expertise. This framework offers both,” he said in the news release.
The researchers highlight that utilizing blockchain reduces costs and risks by distributing computational tasks across numerous machines. This decentralized approach mitigates potential server crashes or malicious attacks, enhancing reliability and security.
“If a centralized server fails, the whole platform goes down,” Alagha added. “Blockchain gives you distribution and transparency. Everything is logged on it, so it is very difficult to tamper with.”
The framework’s dynamic also vastly expedites the often strenuous model training process. Pre-existing models, once developed, can be adapted to fit a user’s specific needs with minor adjustments. For example, a large city might create a model to optimize traffic flow, which smaller cities could then customize without hefty development costs.
This initiative has seen contributions from Hadi Otrok, Shakti Singh and Rabeb Mizouni of Khalifa University in Abu Dhabi, showcasing international collaboration in advancing accessible AI technology.
This breakthrough in AI technology heralds a future where small and medium-sized businesses can compete on a more level playing field, powered by accessible and transparent AI solutions.