Why does Markov Decision Process matter in Reinforcement Learning?

For most learners, the Markov Decision Process(MDP) framework is the first to know when diving into Reinforcement Learning (RL). However, can you explain why it is so important? Why not another framework? In this post, I will explain the advantages of MDP compared to the k-armed bandit problem, another popular RL framework. The post is inspired by an RL specialization offered by University of Alberta and Alberta Machine Intelligence Institute on Coursera. I wrote this post to summarize some of the videos and get a deeper understanding of the specialization.

Read More

How to Model Experience Replay, Batch Learning and Target Networks

If you believe Deep Q-learning is simply a matter of replacing a lookup table with a neural network, you might be in for a rough awakening. Although Deep Q-learning allows handling very large state spaces and complicated non-linear environments, these benefits come at a substantial cost.

Read More

AI’s Olympic Gold-worthy Goal-Driven Systems

Worthy goal

In her article Eyes on the Prize: Understanding the Links Between Perception and Motivation, behavioral scientist Emily Balcetis describes a series of experiments she dubbed ‘eyes on the prize’, which tested a strategy that motivated people to do something that might otherwise look insurmountable. This amounted to people trying to exercise better and were told to look at the distance to a finish line and focus their gaze solely on that line while ignoring everything else.

Read More

Is DeepMind’s new reinforcement learning system a step toward general AI?

Deepmind reinforcement learning

This article is part of our reviews of AI research papers, a series of posts that explore the latest findings in artificial intelligence. One of the key challenges of deep reinforcement learning models—the kind of AI systems that have mastered Go, StarCraft 2, and other games—is their inability to generalize their capabilities beyond their training domain. This limit makes it very hard to apply these systems to real-world settings, where situations are much more complicated and unpredictable than the environments where AI models are trained.

Read More

Reinforcement Learning vs Genetic Algorithm — AI for Simulations

I had to decide, on an optimization approach that would better suit the use case. I had Reinforcement Learning and Genetic Algorithm (the two roads among many) in mind, but then an epiphany… “Both are nature inspired AI approaches, how are the two different? And more importantly, in which scenarios, is one favoured over the other?” And thus today we will be dissecting parts of the thought process behind coming to a decision.

Read More

Deep reinforcement learning helps us master complexity

Deep reinforcement learning—where machines learn by testing the consequences of their actions—is one of the most promising and impactful areas of artificial intelligence. It combines deep neural networks with reinforcement learning, which together can be trained to achieve goals over many steps. It’s a crucial part of self-driving vehicles and industrial robots, which have to navigate complex environments safely and on time.

Read More

The Paladin, the Cleric, and the… Reinforcement Learning?

Data science and artificial intelligence are everywhere. So are video games. It’s no surprise that it was only a matter of time until people started getting creative with combining the two in unique ways. And no, I’m not talking about improving in-game AI (because clearly, Skyrim doesn’t care), and I’m not talking about analyzing game sales either. Today, I want to look at some fun ways that people have used reinforcement learning in the world of gaming, ranging from creating new AI to beating the game using bots.

Read More

Training AI: Reward is not enough

AI training

In a recent paper, the DeepMind team, (Silver et al., 2021) argue that rewards are enough for all kinds of intelligence. Specifically, they argue that “maximizing reward is enough to drive behavior that exhibits most if not all attributes of intelligence.” They argue that simple rewards are all that is needed for agents in rich environments to develop multi-attribute intelligence of the sort needed to achieve artificial general intelligence. This sounds like a bold claim, but, in fact, it is so vague as to be almost meaningless. They support their thesis, not by offering specific evidence, but by repeatedly asserting that reward is enough because the observed solutions to the problems are consistent with the problem having been solved.

Read More

DeepMind’s Reinforcement Learning Framework “Acme”

Acme is a Python-based research framework for reinforcement learning, open sourced by Google’s DeepMind in 2020. It was designed to simplify the development of novel RL agents and accelerate RL research. According to their own statement, Acme is used on a daily basis at DeepMind, which is spearheading research in reinforcement learning and artificial intelligence.

Read More

How AI complicates enterprise risk management

Man standing in the city

Despite the gains artificial intelligence has already brought to the enterprise, there is still much hand-wringing over its potential for unintended consequences. While the headlines tend to focus on AI running amok and destroying all mankind, the practical reality is that current generations of AI are more likely to wreak havoc on business processes — and profits — if not managed properly. But how can you control something that, by its nature, is supposed to act autonomously? And by doing so, won’t the enterprise be hampering the very thing that makes AI such a valuable asset in the workplace?

Read More

RLCard: A Toolkit for Reinforcement Learning in Card Games

RLCard is a toolkit for Reinforcement Learning (RL) in card games. It supports multiple card environments with easy-to-use interfaces for implementing various reinforcement learning and searching algorithms. The goal of RLCard is to bridge reinforcement learning and imperfect information games. RLCard is developed by DATA Lab at Texas A&M University and community contributors.

Read More

[Paper] Yoshua Bengio team designs consciousness-inspired Planning Agent for Model-Based RL

Imagine you’re in an airport, searching for your departure gate. Humans have an excellent ability to extract relevant information from unfamiliar environments to guide us toward a specific goal. This practical conscious processing of information, aka consciousness in the first sense (C1), is achieved by focusing on a small subset of relevant variables from an environment — in the airport scenario we would ignore souvenir shops and so on and focus only on gate-number signage — and it enables us to generalize and adapt well to new situations and to learn new skills or concepts from only limited examples.

Read More

12 active AI Game competitions (ongoing & upcoming)

Rubic's on screen

AI game competitions are also known as AI programming competitions or bot programming competitions. They can be a great place to practice programming, algorithms, and AI/ML. The competitions vary widely in their difficulty, prizes, languages available, and feasible strategies. To help you find the right one, I’ve compiled a list of ongoing and upcoming AI game competitions to check out below (up-to-date as of May 2021).

Read More

A simple outline of Reinforcement Learning

Do you know how machines (or computers) are able to surpass human performance in complex games like chess and go (DeepMind’s AlphaGo, AlphaZero and MuZero), or drive cars without human intervention? The answer is hidden in the way they are programmed. Such machines are programmed to maximize some objective, which is explicitly defined by humans. This approach is called Reinforcement Learning, and completely mimics how humans and animals learns to act in world.

Read More

Quantum Reinforcement Learning, the next step, but why?

Quantum Computer seems to be the next step to improve our Computing hardware (given how compute-demanding AI has become), but its counterintuitive properties also offers many interesting directions to improve SOTA in AI. In this blog, we will discuss its significance to Reinforcement Learning on an intuitive level, without math.

Read More
1 2 3 4