31–40 of 47 posts
Next: - Autosuggest database tables to use - Automatically reserve parallel computing resources - Autodetect data health issues and auto fix them - Autodetect concept drift and auto fix it - Auto engineer features and interactions - Autodetect leakage and fix it - Autodetect unfairness and auto fix it - Autocreate more weakly-labelled training data - Autocreate descriptive statistics and model eval stats - Autocreate…
This reminds me of John Koza’s Genetic Programming, a technique for evolving small programs. There is an old Common Lisp library to play with it.
My reaction too. They've reinvented genetic/evolutionary programming. They should probably read some of the decades of work that have already been done on it.
This reminds me of John Koza’s Genetic Programming, a technique for evolving small programs. There is an old Common Lisp library to play with it.
My reaction too. They've reinvented genetic/evolutionary programming. They should probably read some of the decades of work that have already been done on it.
"An early example of a symbolically discovered optimizer is that of Bengio et al. [8], who represent F as a tree: the leaves are the possible inputs to the optimizer (i.e. the xi above) and the nodes are one of {+, −, ×, ÷}. F is then evolved, making this an example of genetic programming [36]. Our search method is similar to genetic programming but we choose to represent the program as a sequence of instructions—like a programmer would type it—rather than a tree. "
"[36]" is "Koza, J. R. and Koza, J. R. Genetic programming: on the programming of computers by means of natural selection. MIT press, 1992."
AutoML-Zero aims to automatically discover computer programs that can solve machine learning tasks, starting from empty or random programs and using only basic math operations. If this system is not using human bias, who is it choosing what good program is? Surely, human labeling data involves humans adding their bias to the data? It seems like AlphaGoZero was able to do just end-to-end ML because it was able to use…
> It seems like AlphaGoZero was able to do just end-to-end ML because it was able to use a very clear and "objective" standard, whether a program wins or loses at the game of Go. Just a fun note: winning or losing at the game of Go is actually surprisingly subjective: https://en.wikipedia.org/wiki/Go_(game)#Scoring_rules
Btw, if any Chinese player is reading this, how do you count the score while playing? Do you count territory and remember the number of captured stones or do you count both stones and territory? Thanks.
Earlier quoted context omitted.
> It seems like AlphaGoZero was able to do just end-to-end ML because it was able to use a very clear and "objective" standard, whether a program wins or loses at the game of Go. Just a fun note: winning or losing at the game of Go is actually surprisingly subjective: https://en.wikipedia.org/wiki/Go_(game)#Scoring_rules
The game ends by agreement of the players. If they don't agree on the result ("those stones are alive!") they must keep playing. Chinese rules are much better at this than Japanese ones especially (IMHO) the old ones with the group tax. There are no ambiguities there. Unfortunately the group tax is unpleasant and Chinese rules are a pain to score manually. Japanese rules are full of flaws but are such a nice shortcut…
Ambiguities? No. Subjectivity? Yes.
Here is a simple test: get me data to predict the future. Can an algo like this learn to read APIs, build scripts, sign up and pay fees, collect data (laying down a lineage for causal prediction), set up accounts, figure out how account actions work and then take actions profitably without going bust?
If it can even do the first part of this I am in. But I doubt it. This is still just at the level of "cool! Your dog can play mini golf."
Earlier quoted context omitted.
Poor data scientists, now whose heads get cut when things go wrong and companies lose billions?
In the days when Sussman was a novice, Minsky once came to him as he sat hacking at the PDP-6. “What are you doing?”, asked Minsky. “I am training a randomly wired neural net to play Tic-Tac-Toe” Sussman replied. “Why is the net wired randomly?”, asked Minsky. “I do not want it to have any preconceptions of how to play”, Sussman said. Minsky then shut his eyes. “Why do you close your eyes?”, Sussman asked his teacher…
It should be nice using a more informative language for giving advice. If this story is tagged as "popular story for dummies" I would feel we are making real progress.
Just one of Minsky great ideas related to reinforcement learning: The credit assignment problem:How do you distribute credit for success among the many decisions that may have been involved in producing it?, in "Steps Toward Artificial Intelligence" (Minsky, 1961): All of the methods we discuss in this book are, in a sense, directed toward solving this problem.
That book is linked from HN and it has just one comment, so I think that NDNS, no dumb nerd stories, will never become popular.
(1) https://news.ycombinator.com/item?id=10972522
More from (2) Minsky in 1951 built the world's first “randomly wired neural network learning machine,” called the stochastic neural-analog reinforcement computer (snarc)
https://www.geek.com/blurb/marvin-minsky-ai-has-been-brain-d...
A fair paper: Exploring Randomly Wired Neural Network for Image Recognition.
Was Sussan at the edge of envisioning deep learning?, then in fact the room has dissapeared!
Earlier quoted context omitted.
My reaction too. They've reinvented genetic/evolutionary programming. They should probably read some of the decades of work that have already been done on it.
Same here. When I studied genetic programming, I was hoping that's where problem solving evolve from as it was flawless. But recent events prove otherwise which made me believe we are using the wrong tool for the wrong problem. Here is why. When AI gets to 100% accuracy, the equation to find the answer becomes 100% accurate. We no longer have to run the AI with heavy resources and equation can be converted to an exec…
Basic Tech Bros still don't get it. This is cool but real problem is finding/defining the problem. And you don't get a million guesses. Here is a simple test: get me data to predict the future. Can an algo like this learn to read APIs, build scripts, sign up and pay fees, collect data (laying down a lineage for causal prediction), set up accounts, figure out how account actions work and then take actions profitably w…
This is in the site guidelines: https://news.ycombinator.com/newsguidelines.html.
Earlier quoted context omitted.
My reaction too. They've reinvented genetic/evolutionary programming. They should probably read some of the decades of work that have already been done on it.
Same here. When I studied genetic programming, I was hoping that's where problem solving evolve from as it was flawless. But recent events prove otherwise which made me believe we are using the wrong tool for the wrong problem. Here is why. When AI gets to 100% accuracy, the equation to find the answer becomes 100% accurate. We no longer have to run the AI with heavy resources and equation can be converted to an exec…