Machine Learning At Google: The Amazing Use Case Of Becoming A Fully Sustainable Business

Google’s mission is to organize the world’s information and make it universally accessible and useful. From the start, they have also made significant efforts to do this in a way that doesn’t deplete the world’s natural resources.

The company has been fully carbon neutral since 2007 and ten years later they are hoping they have achieved the next major goal – drawing every watt of energy they use for their business operations from renewable sources.

Kate E Brandt, their lead for sustainability, spoke to me about some of the ways they have been tackling this ambitious challenge while she was visiting London to speak at the Economist Sustainability Summit 2018.

She told me “We set a goal in 2012 that we wanted to purchase 100% renewable energy for our operations – so it’s a longstanding commitment.

“We are completing our final calculations but all our indicators point to us having achieved that in 2017 – but stay tuned!”

Of course, Google being pioneers of machine learning and deep learning means they have some formidable technology available to them to achieve this. As you would expect, it has been deployed across a wide variety of use cases in order to achieve their aims.

With data centers accounting for 2% of the world’s global energy usage, creating efficiencies across its own network of 14 major hubs has been a priority for Google.

The challenge here is that the hugely complex nature of the equipment means there are literally billions of possible configurations of servers, chillers, cooling towers, heat exchangers and control systems. Knowing which configurations will lead to the optimum level of Power Usage Effectiveness (PUE) – the metric used by Google to rate energy efficiency in data centers – is insanely complex for human beings to work out. Even a team of highly trained Google data center engineers.

But they took it as far as they could – building their own centers from the ground up so as to have maximum control over the variables at play, and custom-designing components to be free of extraneous, resource-sapping features common in off-the-shelf components.

Google – specifically, one engineer named Jim Gao – then turned to machine learning, the same technology which powers its image recognition and translation applications used by millions worldwide, to take things a step further.

Brandt says “So Jim took a machine learning course online, and got to thinking that it was really an interesting idea for optimizing data center cooling.

“One thing he told me which makes it so powerful as a tool – if you think about 10 devices each of which have 10 settings, that’s 10 billion potential configurations, and not something that the human mind can optimize.

“But once he was able to train this algorithm to see patterns across the various systems and how they impacted the cooling infrastructure, he was able to see that there was a tremendous opportunity.”

Machine learning basically involves feeding complex algorithms, designed carry out data processing tasks in a similar way to the human brain, with huge amounts of data. The result is computer systems which become capable of learning.

 The outcome of a pilot conducted by Jim and his team was a further 40% reduction in the overall amount of energy used for cooling the data center.

“It really shows how we’re able to use this technology across a complex system that’s already highly optimized, and see tremendous results. It highlights what’s so exciting about the potential of machine learning,” Brandt says.

Reducing the amount of waste going into landfill is another environmental priority for Google, and Brandt tells me that currently, they have achieved a “landfill diversion rate” of 86% in their global data centers – meaning just 14% of waste products are not recycled, composted or reused in some way.

This involved taking an aggressive look at every aspect of operation right down to the treatment of food waste from the company restaurants. Google employees are fed three meals a day from 200 cafes and 1,000 self-service eateries. Inevitably this resulted in a certain amount of food going to waste through spoilage or miscalculating demand.

Through a partnership with food data specialists LeanPath, “smart” scales equipped with cameras to precisely measure the amount of food going to waste – either in the kitchen or after being served and left on plates.

All this data is then analyzed to gain an overall understanding of where food is being overproduced and going to waste. The system is credited with cutting the amount of food waste produced by the business by 3 million lbs (pounds) since it was introduced in 2014.

Brandt says “Sustainability provides us with both challenges and opportunities – we are very focused on the idea of the ‘circular economy’ and have really been looking at everything we do as a company, to support a shift in the company and change our relationship with natural resources.

“With AI and machine learning we see tremendous opportunities to unlock new insights relating to sustainability and I think we are seeing some tremendous opportunities emerging to improve the lives of people on the planet.”

Bernard Marr

Source- https://www.forbes.com

Share

Leave a Reply

Be the First to Comment!

Notify of
wpDiscuz
Skip to toolbar