Tomorrow’s workplace can be run on mind-boggling amounts of information. To make sense of all of it, businesses, developers and individuals will need higher artificial intelligence (AI) systems, higher trained AI employees and more efficient number-crunching servers.
While big tech corporations have the resources and expertise to satisfy these demands, they continue to be beyond the reach of most small and medium-sized enterprises and individuals. To reply to this need, a Concordia-led international team of researchers has developed a brand new framework to make complex AI tasks more accessible and transparent to users.
The framework, described in an article published within the journal Information Sciences, makes a speciality of providing solutions to deep reinforcement learning (DRL) requests. DRL is a subset of machine learning that mixes deep learning, which uses layered neural networks to seek out patterns in huge data sets, and reinforcement learning, wherein an agent learns the right way to make decisions by interacting with its environment based on a reward/penalty system.
DRL is utilized in industries as diverse as gaming, robotics, health care and finance.
The framework pairs developers, corporations and individuals which have specific but out-of-reach AI needs with service providers who’ve the resources, expertise and models they require. The service is crowdsourced, built on a blockchain and uses a sensible contract — a contract with a pre-defined set of conditions built into the code — to match the users with the suitable service provider.
“Crowdsourcing the means of training and designing DRL makes the method more transparent and more accessible,” says Ahmed Alagha, a PhD candidate on the Gina Cody School of Engineering and Computer Science and the paper’s lead writer.
“With this framework, anyone can join and construct a history and profile. Based on their expertise, training and rankings, they might be allocated tasks that users are requesting.”
Democratizing DRL
In line with his co-author and thesis supervisor Jamal Bentahar, a professor on the Concordia Institute for Information Systems Engineering, this service opens the potential offered by DRL to a much wider population than was previously available.
“To coach a DRL model, you would like computational resources that aren’t available to everyone. You furthermore may need expertise. This framework offers each,” he says.
The researchers consider that their system’s design will reduce costs and risk by distributing computation efforts via the blockchain. The doubtless catastrophic consequences of a server crash or malicious attack are mitigated by having dozens or a whole lot of other machines working on the identical problem.
“If a centralized server fails, the entire platform goes down,” Alagha explains. “Blockchain gives you distribution and transparency. Every part is logged on it, so it is rather difficult to tamper with.”
The difficult and dear means of training a model to work properly might be shortened by having an existing model available that only requires some relatively minor adjustments to suit a user’s particular needs.
“For example, suppose a big city develops a model that may automate traffic light sequences to optimize traffic flow and minimize accidents. Smaller cities or towns may not have the resources to develop one on their very own, but they will use the one the large city developed and adapt it for their very own circumstances.”
Hadi Otrok, Shakti Singh and Rabeb Mizouni of Khalifa University in Abu Dhabi contributed this study.