Statistical machine learning methods often face the challenge of limited data available from the population of interest. One remedy is to leverage data from auxiliary source populations, which share some conditional distributions or are linked in other ways with the target domain. Techniques leveraging such \emph{dataset shift} conditions are known as \emph{domain adaptation} or \emph{transfer learning}. Despite extensive literature on dataset shift, limited works address how to efficiently use the auxiliary populations to improve the accuracy of risk evaluation for a given machine learning task in the target population. In this paper, we study the general problem of efficiently estimating target population risk under various dataset shift conditions, leveraging semiparametric efficiency theory. We consider a general class of dataset shift conditions, which includes three popular conditions -- covariate, label and concept shift -- as special cases. We allow for partially non-overlapping support between the source and target populations. We develop efficient and multiply robust estimators along with a straightforward specification test of these dataset shift conditions. We also derive efficiency bounds for two other dataset shift conditions, posterior drift and location-scale shift. Simulation studies support the efficiency gains due to leveraging plausible dataset shift conditions.

This article describes one person's experience in coming from an academic environment to work at Digital Equipment Corpo I've divided this history into two distinct parts. AI and DEC's entry into the AI market, DEC engineers were This article is an edited version of Dr Polit's presentation at the Technology Transfer Symposium held at the AAAI-83 conference Building Expert Systems I'll now give a brief review of the steps involved in building expert systems as they are described by many researchers. The five steps involved in building an expert system are: Step 1: problem recognition, Step 2: task definition, Step 3: initial design, Step 4: knowledge acquisition, and Step 5: system maintenance. Frequently, the problem is perceived as a bottleneck in a larger process; sometimes it is a scarcity of traiued personnel. Second, during step 2, researchers must define the functions the AI system will perform.

A Broadened Perspective of Manufacturing: The Knowledge Network In order to form a vision and a strategy, we took a broad new look at our manufacturing business. The perspective ranged from the customer at the point of sale through point of manufacture and point of distribution and back to the customer. In 1981 DEC coined the term knowledge network to represent this notion (O'Connor 1984) (see figure 1). In many of these "pockets of expertise, " within DEC or any other manufacturing business, the expertise and the reasons for making decisions are generally undocumented or are unavailable to all the parties needing the information. Two Views of the Business Within the knowledge network two major cycles are apparent: the order-process cycle and the product life cycle The order-process cycle (see figure 2) is oriented around taking, manufacturing, delivering, and servicing an order.

The roots of modern Artificial Intelligence, or AI, can be traced back to the classical philosophers of Greece, and their efforts to model human thinking as a system of symbols. More recently, in the 1940s, a school of thought called "Connectionism" was developed to study the process of thinking. In 1950, a man named Alan Turing wrote a paper suggesting how to test a "thinking" machine. He believed if a machine could carry on a conversation by way of a teleprinter, imitating a human with no noticeable differences, the machine could be described as thinking. His paper was followed in 1952 by the Hodgkin-Huxley model of the brain as neurons forming an electrical network, with individual neurons firing in all-or-nothing (on/off) pulses.

The rapid advances in information technology are causing a fundamental change in the way we do our business. Within our manufacturing business today, various parts of the organization are " reasoning " about "engineered products." The everyday problem-solving activity within the organization can be thought of as conducted by a network of experts knowledgeable about the products and the physical and paperwork processes that constitute the business, that is, the knowledge network. The focus of our attention has not been just at the factory level; we have been addressing the order-process cycle: marketing, sales, order administration, manufacturing, distribution, and field service. This cycle can be thought of as outer loop of the knowledge network. Also, we recently began addressing the inner loop. This loop is the product life cycle : marketing and new product requirements, design and manufacturing startup, and volume or steady-state manufacturing. This article describes DEC's internal strategy for applying artificial intelligence (AI) to manufacturing processes and problems above the work-cell level. In addition to an overview of this knowledge network, we feature DEC's newest system in order processing : the configuration-dependent sourcing (CDS) expert. Project experience on this system, which deals with the assignment of fulfillment sites (factories) to line items in computer system orders, is also described.

This article describes one person's experience in coming from an academic environment to work at Digital Equipment Corporation. The author feels his own experience has paralleled the transfer of AI technology from academia to industry, where AI researchers must live up to very different expectations, but also enjoy very different rewards. This article covers the historical background of DEC's involvement with AI, the development of R1- known internally and henceforth in this article as XCON-and DEC's experiences with it and its consequences. Finally, the article offers advice for other corporations planning to develop their own capabilities in AI.