Sign in Sign up. Launching GitHub Desktop Go back. Launching Xcode Launching Visual Studio Latest commit 57c0bc3 Sep 12, Contributions most welcome. Course run by Peter Norvig EdX Artificial Intelligence - The course will introduce the basic ideas and techniques underlying the design of intelligent computer systems Artificial Intelligence For Robotics - This class will teach you basic methods in Artificial Intelligence, including: probabilistic inference, planning and search, localization, tracking and control, all with a focus on robotics Machine Learning - Basic machine learning algorithms for supervised and unsupervised learning Neural Networks For Machine Learning - Algorithmic and practical tricks for artifical neural networks.
Stanford Statistical Learning - Introductory course on machine learning focusing on: linear and polynomial regression, logistic regression and linear discriminant analysis; cross-validation and the bootstrap, model selection and regularization methods ridge and lasso ; nonlinear models, splines and generalized additive models; tree-based methods, random forests and boosting; support-vector machines.
Python Class By Google This is a free class for people with a little bit of programming experience who want to learn Python. The class includes written materials, lecture videos, and lots of code exercises to practice Python coding. Deep Learning Crash Course In this liveVideo course, machine learning expert Oliver Zeigermann teaches you the basics of deep learning. Paradigms Of Artificial Intelligence Programming: Case Studies in Common Lisp - Paradigms of AI Programming is the first text to teach advanced Common Lisp techniques in the context of building major AI systems Reinforcement Learning: An Introduction - This introductory textbook on reinforcement learning is targeted toward engineers and scientists in artificial intelligence, operations research, neural networks, and control systems, and we hope it will also be of interest to psychologists and neuroscientists.
The Cambridge Handbook Of Artificial Intelligence - Written for non-specialists, it covers the discipline's foundations, major theories, and principal research areas, plus related topics such as artificial life The Emotion Machine: Commonsense Thinking, Artificial Intelligence, and the Future of the Human Mind - In this mind-expanding book, scientific pioneer Marvin Minsky continues his groundbreaking research, offering a fascinating new model for how our minds work Artificial Intelligence: A New Synthesis - Beginning with elementary reactive agents, Nilsson gradually increases their cognitive horsepower to illustrate the most important and lasting ideas in AI On Intelligence - Hawkins develops a powerful theory of how the human brain works, explaining why computers are not intelligent and how, based on this new theory, we can finally build intelligent machines.
Also audio version available from audible. The Elements of Statistical Learning: Data Mining, Inference, and Prediction - Hastie and Tibshirani cover a broad range of topics, from supervised learning prediction to unsupervised learning including neural networks, support vector machines, classification trees and boostingthe first comprehensive treatment of this topic in any book. Deep Learning and the Game of Go - Deep Learning and the Game of Go teaches you how to apply the power of deep learning to complex human-flavored reasoning tasks by building a Go-playing AI.
After exposing you to the foundations of machine and deep learning, you'll use Python to build a bot and then teach it the rules of the game. Deep Learning for Search - Deep Learning for Search teaches you how to leverage neural networks, NLP, and deep learning techniques to improve search performance. Deep Learning with PyTorch - PyTorch puts these superpowers in your hands, providing a comfortable Python experience that gets you started quickly and then grows with you as you—and your deep learning skills—become more sophisticated.
Deep Learning with PyTorch will make that journey engaging and fun. Grokking Deep Reinforcement Learning - Grokking Deep Reinforcement Learning introduces this powerful machine learning approach, using examples, illustrations, exercises, and crystal-clear teaching. Fusion in Action - Fusion in Action teaches you to build a full-featured data analytics pipeline, including document and data search and distributed data clustering.
The Cambridge Handbook of Artificial Intelligence. Ramsey Edited by Keith Frankish. Publisher: Cambridge University Press , This specific ISBN edition is currently not available. View all copies of this ISBN edition:. Synopsis About this title Artificial intelligence, or AI, is a cross-disciplinary approach to understanding, modeling, and creating intelligence of various forms.
Shop by category
Book Description : This volume of original essays by leading experts describes cutting-edge work in artificial intelligence. Buy New Learn more about this copy. About AbeBooks. Other Popular Editions of the Same Title. Search for all books with this author and title. Customers who bought this item also bought. Stock Image. Published by Cambridge University Press New Paperback Quantity Available: Seller Rating:.
New Hardcover Quantity Available: Book Depository hard to find London, United Kingdom.
- Join Kobo & start eReading today?
- The Pictorial Guide to the Living Primates.
- Fler böcker av författarna?
- A Land of Ash.
- What is Kobo Super Points?.
- Saint Thomas and Platonism: A Study of the Plato and Platonici Texts in the Writings of Saint Thomas.
- The Manga Cookbook.
New Hardcover Quantity Available: 1. The point is that we have little conceptual clarity in the great AI debate because there is little agreement on what to include or exclude. If we cannot find the lines of inclusion and exclusion, then having any serious discussion about the possible benefits and risks of AI is almost impossible. A diagram of a basic neural network used in machine learning. Image credit: Kjell Magne Fauske. It should become apparent that however we decide to define or to limit discussion of what AI is, such as whether it includes GOFAI or is merely machine learning, AI can be applied almost anywhere.
After all, to create and tailor AI for a given situation, we only need four things: computing power, expertise, data, and an algorithm.
The Cambridge Handbook of Artificial Intelligence - Frankish, Keith -
Thus, as society increasingly produces and sheds data, and the capabilities of computer processing and transfer speeds increase, AI will undoubtedly be applied simply because it can be. Most of the necessary algorithms are decades old, open source, and easily accessed. Despite the shifting power balance, there are some ways to democratize these powers even more. This is due to some simple facts about the potential scale and use of AI. In other words, for a problem to be well suited for an AI application, we really need to have a lot of latent and background understanding about that problem, something that is often not appreciated.
To explain, there are some basic notions from control theory that underpin most AI systems. Control theory deals with dynamic systems; that is, systems that interact with their environment. We might control a system by changing the desired task, the required sensors, the data gathered, the information processes, or the guidance or architecture for how the system acts or reacts.
The actions of the system can be few or many, and the choice of what to pursue can be narrow or wide. But what its construction ultimately requires is a robust understanding of what the system needs to function appropriately. For we must not only be able to clearly identify the desired task and construct a system that is capable of carrying it out but also be able to understand how and in what manner the task was completed.
This is where the boundary between development and deployment matters most, because there are all sorts of other metrics that come into play. For instance, did the system act safely? Did it act in accordance with laws, regulations, or norms? Answering these questions allows us to gain a clearer understanding of which problems are well suited for AI and which are more troublesome.
The Cambridge Handbook of Artificial Intelligence: The ethics of artificial intelligence
Indeed, it also allows us to acknowledge that some problems cannot be solved with AI. For tasks that are well defined and narrow, such as calculating the estimated trajectories of an object, an AI agent can do quite well. This is because we understand how gravity, drag, velocity, and the general laws of physics work.
For such problems, even if we do not tell an agent beforehand what these properties are but allow it to learn them through iteration, given sufficient interaction with the environment, it can figure out how to do something even if not understanding why. The context of each skill or task may change, the environment may change, and the reasoning required to accomplish different tasks may be of a much higher order.
This is mostly due to the fact that transferring how an agent perceives, reasons about, or finds patterns within its input data to another domain or task is extremely difficult, often causing the entire system to fail. It is no surprise, then, that many of the areas where AI is seen as being most beneficial are areas with large amounts of data to draw on, where the structure of the task is generally known, and where there is a way to easily check whether it has correctly and appropriately completed its task. For example, limited forms of medical diagnostics and natural language processing have both seen major advancements thanks to AI.
However, many of the applications that scholars and policymakers have proposed for AI are embedded in social and political structures, and the decisions and actions that an AI system takes will affect those structures. If one believes that such surveillance and the limitation of particular freedoms is a violation of human rights, then that person might argue that the mere availability of AI to enable those rights violations could further entrench beliefs about particular classes of people, could give rise to new vulnerable populations, and could even lead to destabilizing conditions in the international system.
As I have hinted at, many of the questions in the AI debate are actually epistemological questions, and they are not confined to AI. Rather, they are age-old questions about whether we have some ground truths from which even to proceed. Epistemology, or a theory of knowledge, for an AI is necessary because the notion of an AI is premised on the derivative concept of intelligence.
A theory of knowledge, with its attendant requirement of offering differing types of knowledge and standards for justified belief, validity, error, and the like, is required if we are to make claims about whether the system is acting intelligently and appropriately. Additionally, what constitutes knowledge for an AI agent is tightly coupled with such things as what sorts of information it can access, identify, and assess; the formal relation between its capacity to reason, its sensory abilities, and the decisions it makes; and its ability to update and adapt based on interactions with its environment, other agents human users or other AI agents , or even simulated data.
The lower the correspondence, the greater the error. But to once again reiterate: If we are to build systems that are capable of percent knowledge correspondence, then we first must understand what objective reality is or whether it even exists. This may seem like a truism but, in effect, we must acknowledge that humans live in a complex web of social interactions, norms, customs, and power relations.
There may be some form of intersubjective objectivity—that is, we all subjectively agree on what constitutes reality and truth—but the basis of this is purely a subjective feeling, experience, or thought. Why does this matter for understanding the future of AI? It matters on two fronts. The first is related to whether we can, from a technological standpoint, create AI systems to carry out tasks related to the subjectivities noted above.
Here, we are presented with questions about the type and quality of data given to train an agent. For instance, I may be able to train an autonomous vehicle to drive on the left versus the right side of the road, but I may not be able to train an AI to identify who is a combatant in war. I may be able to train an AI agent to identify certain types of weapons or even people, but I currently cannot train it to understand the more fluid and dynamic notion of combatancy. This is because combatancy is a behavior that is dependent upon hostile intent , and there are myriad forms that it may take in a variety of environments, with almost infinite kinds of weapons.
- Cambridge Handbook of Artificial Intelligence.
- The Cambridge Handbook of Intelligence - PDF Free Download.
- Blood and Guts.
- From Me to We: Why Commercial Collaboration Will Future-proof Business, Leaders and Personal Success.
- The Cambridge Handbook of Artificial Intelligence | Cokesbury.
Thus, when discussing the morality or potential efficacy of autonomous weapons systems, one must first determine whether the system is capable of attacking particular kinds of targets and how it came to those determinations.
Related The Cambridge Handbook of Artificial Intelligence
Copyright 2019 - All Right Reserved