Radar Telegram alerts Open Radar

Richard W. Hamming, The Art of Doing Science and Engineering (1997) A great book on how to think as a scientist and engineer, derived from Hamming's lecture course (the one for which codes and distance are called). Along the way, it's the history of computing through the eyes of someone who watched it almost from the beginning, as well as coding and information theory, digital filters, simulations, and three chapters on AI. Apparently everyone knows, but again I will recommend the chapter on one-dimensional space, where on the fingers he shows how our three-dimensional intuition breaks down in high dimensions: the volume of the sphere goes to the surface, and the count between the spheres at the corners of the cube, starting with n = 10, climbs beyond its edge. The ML-man is a great reader. Hamming's experience is very relevant right now. It collected high-level languages ​​when experienced programmers believed that only low-level writing could be reliable and efficient. The same is happening with AI right now, you've probably heard all this argument about why code written by agents won't work. Combined with this story is a classic from Rich Sutton The Bitter Lesson about how over 70 years of AI research, common methods at scale along with computing, namely, search and learning, repeatedly beat the systems where researchers manually put their knowledge to the task. Ling also has a magical lecture on how to make rice, you and your research, which is the last chapter of the book. And a good paper, The Unreasonable Effectiveness of Mathematics (1980), is a response to Weiser's famous paper, with an unusual thought: If there are smells and sounds that dogs can hear but cannot see, why not thoughts that we cannot think? Evolution may close us to certain lines of thought. 5 Clay M. Christensen, The Innovator Dilemma (1997) An old but important book on disruptive innovation.

תרגם לעברית

gonzo ML papers · · Source

Richard W. Hamming, The Art of Doing Science and Engineering (1997) A great book on how to think as a scientist and engineer, derived from Hamming's lecture course (the one for which codes and distance are called). Along the way, it's the history of computing through the eyes of someone who watched it almost from the beginning, as well as coding and information theory, digital filters, simulations, and three chapters on AI. Apparently everyone knows, but again I will recommend the chapter on one-dimensional space, where on the fingers he shows how our three-dimensional intuition breaks down in high dimensions: the volume of the sphere goes to the surface, and the count between the spheres at the corners of the cube, starting with n = 10, climbs beyond its edge. The ML-man is a great reader. Hamming's experience is very relevant right now. It collected high-level languages ​​when experienced programmers believed that only low-level writing could be reliable and efficient. The same is happening with AI right now, you've probably heard all this argument about why code written by agents won't work. Combined with this story is a classic from Rich Sutton The Bitter Lesson about how over 70 years of AI research, common methods at scale along with computing, namely, search and learning, repeatedly beat the systems where researchers manually put their knowledge to the task. Ling also has a magical lecture on how to make rice, you and your research, which is the last chapter of the book. And a good paper, The Unreasonable Effectiveness of Mathematics (1980), is a response to Weiser's famous paper, with an unusual thought: If there are smells and sounds that dogs can hear but cannot see, why not thoughts that we cannot think? Evolution may close us to certain lines of thought. 5 Clay M. Christensen, The Innovator Dilemma (1997) An old but important book on disruptive innovation.

More on Radar

Want to see every report in real time?

Hundreds of sources worldwide, translated into your language, with alerts for what matters to you.

Sign up free
Share on WhatsApp Get alerts on Telegram