Promotions - Choice's Outstanding Academic Titles 2025 -

From his own experiences and contributions to the field, the late Eugene Charniak (Brown Univ., Professor Emeritus of Computer Science and Cognitive Science) traces the intellectual history of artificial intelligence (AI) from 1956 through 2023. Chapters are organized by topics and periods such as "Computer Vision 1970 - 2000", "Learning Language 1985 - 2010" and "Deep Learning 1989 - 2016". Charniak introduces and frequently reiterates contrasting concepts such as statistical and probabilistic methods, and problem solving and reinforcement learning. Charniak devotes much coverage to rule making, training, and learning including in perfect information games such as checkers, chess, and Go. The density of concepts and model examples are surprisingly detailed for a short work, which may make for challenging reading, especially for the non-scientist or practitioner. However, frequent illustrations are helpful, and overall this book demystifies the often amorphous topic of AI. In contrast to unfounded claims of the economic benefits or societal dangers of AI, Charniak presents a realistic vision of AI based upon research and development including the many disappointments in machine approximation of human senses and learning. The book includes a timeline and bibliography.Summing Up: Highly recommended. Upper-division undergraduates through faculty; professionals.

Policy wonks acknowledge that regulation can't keep up with changes in technology, but they rarely offer solutions. Holt (Univ. of California, Santa Barbara) offers suggestions in this analysis of pipelines, platforms, and data that make up "the cloud." She's quick to point out that use of the cloud metaphor (rather than "closet" or "storage locker") obscures the infrastructure needed to support it. The book asserts that regulation of such an essential service has been lacking. "Internet access must be viewed as an essential public service which requires policy rationales and foundations for broadband provision to match that of other utilities" (p. 29). The book does an outstanding job of examining how we got where we are. Pipelines, platforms, and data are each examined separately, and the book provides specific recommendations for ensuring a cloud that serves all. Most prophetic of all is the admonition, "Reforming cloud policy in the public interest will require the public's interest" (p. 199). Some readers may not agree with her conclusions, but there is no arguing that this book is well-researched and addresses a crucial issue at an important time.Summing Up: Highly recommended. All readers.

With an elegant brevity, Data Mining and Python provides an overview of the data science lifecycle in Python by providing examples of how to do it while still covering the why to do it. The book covers data preparation--from data collection to data preprocessing and data analysis--covering three major analysis types (classification, regression, and clustering), as well as pattern finding and outlier analysis. The author provides a brief framework for each chapter and then provides a comprehensive example of its use in Python. This makes the book an excellent reference guide, as each section provides a working example, covering data ingest, wrangling, analysis, and parameter tuning. These examples provide the complete Python code and the complete results improving its value for all levels of reader. As the data analysis chapters build on the data wrangling chapters, with a brief introduction to Python, this book could also serve as a textbook for a semester-long data science course.Summing Up: Highly recommended. Lower-division undergraduates through professionals.