FISH FARMING TECHNOLOGY A few years ago, it was cloud, followed by the Internet of things; then blockchain took the centre stage; it seems like there’s always a new buzzword coming from the tech industry. This year, artificial intelligence and machine learning are everywhere, and aquaculture was not spared. The industry is taken by storm, but is it really worth the hype?
could support farmers. For animal health, models can be trained on thousands of images of a specific symptom, and be used to detect disease before the human eye can reliably do so, giving more options to farmers than just harvesting when it’s too late. From there, we can build on complexity, capturing Samuel Couturetreatments made to the pond, measuring the effectiveness of inputs, and adding Brochu variables such as feed usage, broodstock, Chief Technology Officer, XpertSea and water quality. For clarification purposes, machine On the feed side, much optimisation can be done. AI can learning is an application of artificial intelligence (AI) that undoubtedly support biomass estimation, which is at the core provides systems with the ability to automatically learn of feed prescriptions. But imagine adding other variables in the and improve from experience without being explicitly equation, such as water quality and temperature, forecasted rain, programmed. Let’s take a real-life example to demonstrate its application: say you have a young child, and you want to teach density, genetics line, and other factors. It is not far-fetched to imagine a scenario where a farmer him colours. receives an alert on his mobile phone to reduce his feed by 10 First, you might want to teach him to recognise a blue car percent for the next day due to changes in water parameters from a red one. Once he has achieved that, you could show him because of an upcoming rainstorm. There is no doubt AI will other objects that are blue or red. By increasing the number support farmers in making better decisions for their feed which of examples of coloured objects, his brain will make sense of will lead to better feed conversion ratios, and ultimately, more the colour concept, and soon learn that colours are valid for a profit. panoply of objects. Machine learning is very similar, but with While this all looks impressive, AI and machine learning are computers. as good as the data they are built upon. We like to say: garbage In other industries like human health, AI is improving disease diagnostics and doing so with greater accuracy than ever before. in, garbage out. Indeed, if you build a model with inaccurate and improperly labelled data, or simply an insufficient amount As an example, a recent study1 demonstrated that machine of data, its predictions might well be far from reality. learning, in this case, deep learning, is more accurate at For this reason, one of the first questions to ask a tech detecting lung tumours than radiologists. company doing machine learning should be: what is your This is only the beginning. By training AI models on an underlying dataset? There are many aquatech start-ups that increasingly larger number of medical images, it is conceivable market themselves as AI-driven platforms, but some of those that those models would be able to detect cancer much earlier. might be simple rule-based systems like “if this happens, Should radiologists be scared of such a technology? Not at all. then do that”. An AI-driven platform is built on solid data It will support their highly technical work and facilitate foundations and strong machine learning engineering decision making. In this situation, an AI could only show the knowledge, not only on heuristic rules. relevant cases to the expert. It could then suggest a diagnosis AI is revolutionising the aquaculture industry, and we are still and let humans approve it, saving humans time while in the early days. It’s a burgeoning ecosystem of start-ups and empowering radiologists. established companies, tackling the industry’s largest problems Similarly, AI will not replace farmers in aquaculture; it will using AI, and Xpertsea is certainly proud of its current and become a key asset in their day-to-day operations. Disease and future contribution. feed management are two of the most important areas where AI
Aquaculture Without Frontiers (AwF) is a Charitable Incorporated Organisation (CIO) that promotes and supports responsible and sustainable aquaculture and the alleviation of poverty by improving livelihoods in developing countries.
Registered charity No. 1165727
www.aquaculturewithoutfrontiers.org.uk