Scholarlink

According to a recent estimate (Reinsel et al. 2017), by the end of 2025 we will be approaching 200 zettabytes of data in the universe’s database. One zettabyte equals 1021 bytes, or one thousand billion billion bytes. The twenty-first century will therefore be remembered as the era of zettabytes. (Interestingly, 1021 is also the estimated number of synapses in the human brain, as we saw in the previous chapter.) If we were to sit comfortably on the couch and watch an HD video containing 200 zettabytes of data, we would need to wait… 40 trillion years before reaching the end of the projection! 200 zettabytes would be enough to fill the memory of 1.5 trillion iPhones or to record 2500 times all the words ever written by every human being. As illustrated in Fig. 5.1, in 1986, when data began to be counted, there were 280 trillion bytes. In less than 40 years, that number has multiplied by nearly a million times! This corresponds to an average growth rate where the volume of data doubles approximately every 18 months. Interestingly, it follows the same dynamics as Moore’s Law for the increasing computational power of computers! To further illustrate these production rates, a recent analysis estimates that the global internet population has grown from 2.1 billion to 5.2 billion in just a decade (2013–2023). This surge generates unimaginable amounts of new data every minute of every single day, as shown in Fig. 5.2. The explosive growth of data is driven by several factors. IoT devices alone are expected to generate 90 zettabytes of data per year by 2025. Almost 30% of the global data sphere will consist of real-time data. By the end of 2025, every connected person in the world, approximately 75% of the global population, will interact with data over 4900 times/day, which means once every 18 s. These trends highlight that the data we generate is not only increasing in volume but also in complexity, frequency, and immediacy. This poses a challenge that requires sophisticated approaches to store, process, and, most importantly, use data to extract actionable insights.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

What Data for Learning, and What Kind of Learning?

  • 章节

  • 发表日期：

    2025年

查看全文
  • Alfio Quarteroni

摘要

According to a recent estimate (Reinsel et al. 2017), by the end of 2025 we will be approaching 200 zettabytes of data in the universe’s database. One zettabyte equals 1021 bytes, or one thousand billion billion bytes. The twenty-first century will therefore be remembered as the era of zettabytes. (Interestingly, 1021 is also the estimated number of synapses in the human brain, as we saw in the previous chapter.) If we were to sit comfortably on the couch and watch an HD video containing 200 zettabytes of data, we would need to wait… 40 trillion years before reaching the end of the projection! 200 zettabytes would be enough to fill the memory of 1.5 trillion iPhones or to record 2500 times all the words ever written by every human being. As illustrated in Fig. 5.1, in 1986, when data began to be counted, there were 280 trillion bytes. In less than 40 years, that number has multiplied by nearly a million times! This corresponds to an average growth rate where the volume of data doubles approximately every 18 months. Interestingly, it follows the same dynamics as Moore’s Law for the increasing computational power of computers! To further illustrate these production rates, a recent analysis estimates that the global internet population has grown from 2.1 billion to 5.2 billion in just a decade (2013–2023). This surge generates unimaginable amounts of new data every minute of every single day, as shown in Fig. 5.2. The explosive growth of data is driven by several factors. IoT devices alone are expected to generate 90 zettabytes of data per year by 2025. Almost 30% of the global data sphere will consist of real-time data. By the end of 2025, every connected person in the world, approximately 75% of the global population, will interact with data over 4900 times/day, which means once every 18 s. These trends highlight that the data we generate is not only increasing in volume but also in complexity, frequency, and immediacy. This poses a challenge that requires sophisticated approaches to store, process, and, most importantly, use data to extract actionable insights.