Search

This work applied reinforcement learning and the temporal difference TD(λ) algorithm to train a neural network to play the game of Chung Toi, a challenging variant of Tic-Tac-Toe. The effects of changing parameters and settings of the TD(λ) and of the neural network were evaluated by observing the ability of the network to learn the game of Chung Toi and play against a ‘smart’ random player. This work applied techniques that have proven effective in training neural networks in general to the TD(λ) algorithm. The basic implementation of the TD(λ) method resulted in stable performance and achieved a maximal performance of winning 90.4% of evaluation games. When changing parameter settings, the best performance was achieved by using different learning rates between layers in the neural network (92.6% wins), and this was followed by using a relatively high probability of action exploitation (91.8% wins).

We introduce Augmented Efficient BackProp as a strategy for applying the backpropagation algorithm to deep autoencoders, i.e., autoassociators with many hidden layers, without relying on a weight initialization using restricted Boltzmann machines. This training method is an extension of Efficient BackProp, first proposed by LeCun et al. [1], and is benchmarked on three different types of application datasets.

The importance of technology in firm strategy is now well established. Yet a comprehensive opportunity evaluation process that objectively incorporates the multiple technology management aspects remains elusive. The majority of models or tools that do include technological dimensions mainly do so from a normative, point in time, market based perspective. Both strategic researchers and those researchers focused on innovation and technology management have developed tools that are either limited by their age or the scope of their respective field's knowledge basis. The result is a set opportunity evaluation tools that are ineffectual. If this is true then there cause for concern. Established as well as entrepreneurial firm opportunity evaluation processes are dominated by opportunity recognition techniques and respective histories of success in taking any opportunity to market. These processes do not focus on decomposition or objective strategic and technological aspects. This has led to serious problems and spectacular failures in the commercialization of opportunities by entrepreneurs, small and medium sized enterprises, and large firms. Here the authors address this problem. We employ seminal and recent advances in the fields of management of technology, strategy and entrepreneurship to produce a multi-dimensional decomposition based model for opportunity evaluation. We utilize the case study method to verify the model. The resultant is Strategy-Technology Firm Fit Audit model. The model provides established firm decision makers a resource for the investigation of potential opportunities. It also provides an entrepreneurial team, institutional investor, angel or venture capitalists a manner to investigate potential venture success.

This paper presents an extensive evaluation of reservoir computing for the case of classification problems that do not depend on time.We discuss how it is possible to adapt the reservoir approach to learning for the case of static classification problems. Then we present a set of experiments against K-PLS, MLP with entropic cost function and LS-SVM showing that this approach is quite competitive and has the advantage of having only one parameter to be chosen.

This papers considers the relationship between product and process innovation for materials based technologies. This is important since many sources of future products are based on materials innovation these include MEMS and nanotechnologies. Consequently, understanding the relation between product and process innovation is important.

The relationship between leadership style of all managers and the establishment of a quality environment in an R&D setting is considered in this empirical study of 511 engineers. It is found that both transformational leadership and transactional contingent-reward leadership are related to the establishment of a quality environment in the R&D part of a telecommunications firm. However, the impact of transactional contingent-reward leadership ceases to be significant once both leadership styles are considered simultaneously using structural equations. A transformational leadership style is also found to be related to employee satisfaction.

Forecasting of the market size of MEMS is considered. A forecast from 2003 to 2025 is offered-based on a superior method to the forecast techniques used in many of the studies to date. The challenges of forecasting a disruptive innovation such as MEMs are also discussed.

Many important products and technologies were developed in federal laboratories and were driven initially by national needs and for federal applications. For example, the clean room technology that enhanced the growth of the semiconductor industry was developed at Sandia (SNL) decades ago. Laboratories advances in micro-electro-mechanical-systems (MEMS)-an important set of process technologies vital for product miniturization-are occurring at SNL. Each of the more than 500 federal laboratories in the US, are sources of R&D that contribute to America's economic vitality, productivity growth and, technological innovation. However, only a fraction of the science and technology available at the federal laboratories is being utilized by industry. Also, federal laboratories have not been applying all the business development processes necessary to work effectively with industry in technology commercialization. This paper addresses important factors that federal laboratories, federal agencies, and industry must address to translate these under-utilized technologies into profitable products in the industrial sector.

A growing body of evidence suggests that distinctive technological competence provide the basis for superior firm performance. However, few, if any, firms possess all the prerequisite competencies required for a given technology-product-market paradigm as they enter that market. This paper explores the different ways in which existing firms develop and intensify competence in response to the changing requirements of industry standard products over time. This paper discovers that technology is an important factor in determining which learning strategies should be employed for competence development.

A method for the analysis, ranking and selection of R&D projects from a portfolio is outlined and demonstrated. It is proposed that an objective multi-criteria decision making method be used to split a portfolio of projects into accept, consider further and reject sub-groups. Next, 'the consider further' group is examined using a subjective method, the value creation model. Such an approach allows for obvious decisions to be automated and complex decisions to be given careful consideration, an approach that is more consistent with how managers commonly make select/reject decisions. The relation between research strategy and consideration of categorical data is considered in relation to the research portfolio of the Advanced Technology Division of Bell Laboratories.