Problem Solving


Using Uninformed & Informed Search Algorithms to Solve 8-Puzzle (n-Puzzle) in Python

@machinelearnbot

For any such board, the empty space may be legally swapped with any tile horizontally or vertically adjacent to it. Given an initial state of the board, the combinatorial search problem is to find a sequence of moves that transitions this state to the goal state; that is, the configuration with all tiles arranged in ascending order 0,1,…,n 2 1. The search space is the set of all possible states reachable from the initial state. Thus, the total cost of path is equal to the number of moves made from the initial state to the goal state.


Deep Learning, Generative Adversarial Networks & Boxing – Toward a Fundamental Understanding

@machinelearnbot

Let's break down a GAN into its basic components: The overall goal of a standard GAN is to train a generator that generates diverse data samples from the true data distribution, leading to a discriminator that can only classify images as real/generated with a 50/50 guess. In the process of training this network, both the generator and the discriminator learn powerful, hierarchical representations of the underlying data that can then transfer to a variety of specific tasks like classification, segmentation, etc… and use-cases. Now that we have a fundamental understanding of GANs, let's revisit their purpose: to learn powerful representations from unlabelled data (i.e. After training a GAN, most current methods use the discriminator as a base model for transfer learning and the fine-tuning of a production model, or the generator as a source of data that is used to train a production model.


Facebook Research just published an awesome paper on learning hierarchical representations

#artificialintelligence

Suppose we have the data of a social network, now when we embed the nodes and by some distance metric we see that two nodes(users) have a small distance then we can suggest the users to be friends on social network. And they used Poincare ball model for the model of hyperbolic space because it is well suited for gradient based optimization, hence backpropagation. We can see that what this model achieves in a 5 dimensional space is better than what euclidean model achieves in 200 dimensional space. They have used hyperbolic space to embed nodes of hierarchical data and achieved some super awesome results.


Artificial Intelligence: Made Easy w/ Ruby Programming; Learn to Create your * Problem Solving * Algorithms! TODAY! w/ Machine Learning & Data ... engineering, r programming, iOS development): Code Well Academy: 9781530826865: Amazon.com: Books

#artificialintelligence

Well... it's not a bad book but I don't think it covers "AI" in more than a very generic sense. The author introduces a few simple techniques for resolving schedules via brute force iterative approach (not inappropriate for many use cases), simple case logic, and a few tree/acyclic graph solvers. Ruby code is given for all approaches and it's easy to follow. If you're not in that category, I'd skip it and pick up a good general algorithm text covering data structures and searching.


Microsoft Develops Algorithm to "Divide and Conquer" Ms. Pac-Man

#artificialintelligence

Researchers at Microsoft developed an artificial intelligence (AI) algorithm that can achieve the maximum score on Ms. Pac-Man, 999,999, four times greater than the highest human score. The system, according to Microsoft's blog, was developed by a Maluuba, a deep learning startup company which was acquired by Microsoft earlier in the year. The divide-and-conquer method assigns individual AI agents different tasks but also allows them to work together collaboratively through a "top manager." Potential applications include helping a company's sales team make predictions about which customers to target depending on factors such as which clients are up for contract renewal, which contracts are most valuable to the company, and if the customer is available at a particular day or time.


Artificial Intelligence and Problem Solving: Danny Kopec, Christopher Pileggi, David Ungar, Shweta Shetty: 9781944534585: Amazon.com: Books

#artificialintelligence

Danny Kopec taught at Brooklyn College and the CUNY Graduate Center. He authored several books, conference and journal articles, and was an International Chess Master. Christopher Pileggi holds a degree in Computer Information Science and is employed by the Center for Economic Workforce & Development. David Ungar holds a degree in Computer Information Science.


Divide and conquer: How Microsoft researchers used AI to master Ms. Pac-Man - Next at Microsoft

#artificialintelligence

Microsoft researchers have created an artificial intelligence-based system that learned how to get the maximum score on the addictive 1980s video game Ms. Pac-Man, using a divide-and-conquer method that could have broad implications for teaching AI agents to do complex tasks that augment human capabilities. The team from Maluuba, a Canadian deep learning startup acquired by Microsoft earlier this year, used a branch of AI called reinforcement learning to play the Atari 2600 version of Ms. Pac-Man perfectly. She said that's similar to some theories of how the brain works, and it could have broad implications for teaching AIs to do complex tasks with limited information. It may seem strange that it takes some of the most advanced AI research methods to beat something as seemingly simple as a 1980s Atari game.


Artificial Intelligence: Structures and Strategies for Complex Problem Solving (6th Edition): George F. Luger: 9780321545893: Amazon.com: Books

#artificialintelligence

This accessible, comprehensive book captures the essence of artificial intelligence -- solving the complex problems that arise wherever computer technology is applied. With his signature enthusiasm, George Luger demonstrates numerous techniques and strategies for addressing the many challenges facing computer scientists today. George Luger is currently a Professor of Computer Science, Linguistics, and Psychology at the University of New Mexico. He received his Ph.D. from the University of Pennsylvania and spent five years researching and teaching at the Department of Artificial Intelligence at the University of Edinburgh.


ZAP your own brain

FOX News

No, its not the worst writers room scenario ever, but an experiment into brain stimulation carried out by researchers at the U.K.s Queen Mary University of London and Goldsmiths University of London. When the transcranial direct-current stimulation (tDCS) technique was used to suppress a key part of the frontal brain called the left dorsolateral prefrontal cortex, participants were shown to get better at carrying out creative tasks involving out-of-the-box thinking. However, since the left dorsolateral prefrontal cortex is a part of the brain heavily involved in much of our reasoning processes, they got worse at solving problems in which many items needed to be held in mind at once. These patients were more likely to solve hard think out of the box problems, which was remarkable evidence that this brain region might hinder creative problem solving.


The Job Market – Human and Robots – Who Takes What?

#artificialintelligence

He has developed technologies to help in diabetes, blood diseases, cardiovascular disease, and cancer – products based on these technologies are currently saving patients' lives daily. He has won numerous awards, such as the North Carolina Governor's Award for Outstanding Entrepreneurial Contribution and the North Carolina Biomedical Entrepreneur of the Year award. Dr. Oberhardt is the author of a book published in 2013, titled: "Dragonfly Thinking, Problem Solving for a Successful Future". Dr. Oberhardt recently conducted an Apprenticeship Program in Problem Solving as a volunteer for Citizen Schools, a multi-state nonprofit organization focused on helping students in public middle schools in low income communities discover and achieve their dreams.