When multiple agents collaborate to perform distributed operations, they can be modeled as cooperative games. Considering a network of agents work together and they can only communicate in a limited way (e.g., only to neighbor peers), the goal is to maximize the cooperation success globally, or maximize the total value and social welfare of the whole network. The type of cooperation is challenging since the game is not zero-sum. There are not any outside agents to serve as referees. The objective functions may be non-stationary and non-convex. In this paper, each agent is modeled as a content supplier or consumer. Each agent optimizes its own objective locally. We show that each agent self-organizes or converges to its "value" via the principles of quantum computing and game theories. We prove two theorems that can optimize an agent's own objective and simultaneously optimize the global social welfare of its peer network. The quantum intelligence game algorithms are unsupervised and self-organizing, where the weights expressed in quantum neural networks or transformers can be computed from a natural mechanism known as a quantum adiabatic evolution.

A complex enterprise includes multiple subsystems and organizations. The U.S. Marine Corps (USMC) maintenance and supply chain is a complex enterprise and exemplifies a socio-technological infrastructures. It is imperative for the USMC to adopt more advanced data sciences including ML/AI techniques to the entire spectrum or end-to-end (E2E) logistic planning as a complex enterprise including maintenance, supply, transportation, health services, general engineering, and finance. In this paper, we first review an overall framework of leveraging artificial Intelligence to learn, optimize, and win (LAILOW) for a complex enterprise, and then show how a LAILOW framework is applied to the USMC maintenance and supply chain data as a use case. We also compare various machine learning (ML) algorithms such as supervised machine learning/predictive models and unsupervised machine learning algorithms such as lexical link analysis (LLA). The contribution of the paper is that LLA computes stable and sensitive components of a complex system with respective to a perturbation. LLA allows to discover and search for associations, predict probability of demand and fail rates, prepare spare parts, and improve operational availability and readiness.

We demonstrate a machine learning method, namely lexical link analysis (LLA), which can be used to discover high-value information from financial data. LLA is an unsupervised learning method that does not require manually labeled training data. We also demonstrate how to form LLA in a game-theoretic framework. We show that with game theory: high-value information selected by LLA reaches a Nash equilibrium by superpositioning popular and anomalous information and at the same time generates high social welfare, therefore containing higher intrinsic value. We show the results of LLA of two sets of financial data validating and correlating with the ground truth.

We demonstrate a machine learning and artificial intelligence method, i.e., lexical link analysis (LLA) to discover high-value information from big data. In this paper, high-value information refers to the information that has the potential to grow its value over time. LLA is a unsupervised learning method that does not require manually labeled training data. New value metrics are defined based on a game-theoretic framework for LLA. In this paper, we show the value metrics generated from LLA in a use case of analyzing business news. We show the results from LLA are validated and correlated with the ground truth. We show that by using game theory, the high-value information selected by LLA reaches a Nash equilibrium by superpositioning popular and anomalous information, and at the same time generates high social welfare, therefore, contains higher intrinsic value.

We demonstrated a machine learning and artificial intelligence method, i.e., lexical link analysis (LLA) to discover different layers of semantic network that contribute to innovative ideas from big data. The LLA is an unsupervised machine learning paradigm that does not require manually labeled training data. Multilayer value metrics are defined based on game theory for LLA. We showed the following results: 1) the value metrics generated from LLA in a use case of an internet game and crowd-sourcing; 2) the results from LLA are validated and correlated with the ground truth; 3) the game-theoretic LLA can help an information provider to present the information in the most valuable way. The information presentation can solve a problem (e.g., a search request of innovation) that no other information providers can solve (i.e., expertise). In addition, it ties also to a broader context that the unique value can propagate through the consensus. Based on the game-theoretic LLA, an information provider should not always present expertise content or authoritative content but rather with a mixed strategy where each type of content is presented with certain probabilities for the best value overall.

We will demonstrate a distributed recursive method, i.e., Lexical Link Analysis (LLA) and an infrastructure, i.e., Collaborative Learning Agents (CLA) to discover high-value information. The combined system is a unified methodology of discovering high-value information from structured and unstructured heterogeneous data sources. We will demonstrate the LLA/CLA system using a crowdsourcing data source and show how it can be used to discover new knowledge for a widening range of applications and heterogeneous data types.

KEOD 2017 - 9th International Conference on Knowledge Engineering and Ontology Development

We use a hidden Markov model (HMM) based continuous speech recognition system to perform off-line character recognition (OCR) of Arabic printed text. The HMM trainer and recognizer are used without change, however we modify the feature extraction stage to compute features relevant to OCR. Although we begin by segmenting the page into a collection of lines, no further segmentation is necessary for either recognition or training. Experiments on the ARPA Arabic data corpus yield a range of character error rates from under one percent for a single computer font to 2.8% for multiple-font recognition of a wide range of material from books, magazines and newspapers.

In this paper we show how continuous speech recognition methods can be used for character recognition resulting in a technology that is language independent and does not require presegmentation of the data at the character and word levels. In multifont experiments on the ARPA Arabic OCR Corpus an average character error rate of 1.9% is obtained using the BBN BYBLOS continuous speech recognition system with no modifications. A first experiment using the identical system and procedures, trained and tested on a subset of the English Univ. of Washington OCR corpus resulted in 1.4% character error.

This paper examines the implementation of projection pursuit regression (PPR) in the context of machine learning and neural networks. We propose a parametric PPR with direct training which achieves improved training speed and accuracy when compared with nonparametric PPR. Analysis and simulations are done for heuristics to choose good initial projection directions. A comparison of a projection pursuit learning network with a single hidden-layer sigmoidal neural network shows why grouping hidden units in a projection pursuit learning network is useful. Learning robot arm inverse dynamics is used as an example problem.

We incorporate the hierarchical mixtures of experts (HME) method of probability estimation, developed by Jordan (see Neural Computation, 1994), into an HMM-based continuous speech recognition system. The resulting system can be thought of as a continuous-density HMM system, but instead of using Gaussian mixtures, the HME system employs a large set of hierarchically organized but relatively small neural networks to perform the probability density estimation. The hierarchical structure is reminiscent of a decision tree except for two important differences: each "expert" or neural net performs a "soft" decision rather than a hard decision, and, unlike ordinary decision trees, the parameters of all the neural nets in the HME are automatically trainable using the EM algorithm. We report results on the ARPA 5,000-word and 40,000-word Wall Street Journal corpus using HME models.

In this paper, we incorporate the Hierarchical Mixtures of Experts (HME) method of probability estimation, developed by Jordan [1], into an HMM(cid:173) based continuous speech recognition system. The resulting system can be thought of as a continuous-density HMM system, but instead of using gaussian mixtures, the HME system employs a large set of hierarchically organized but relatively small neural networks to perform the probability density estimation. The hierarchical structure is reminiscent of a decision tree except for two important differences: each "expert" or neural net performs a "soft" decision rather than a hard decision, and, unlike ordinary decision trees, the parameters of all the neural nets in the HME are automatically trainable using the EM algorithm. We report results on the ARPA 5,OOO-word and 4O,OOO-word Wall Street Journal corpus using HME models.

The current state-of-the-art in large-vocabulary, continuous speech recognition is based on the use of hidden Markov models (HMM). In an attempt to improve over HMM performance, the authors developed a hybrid system that combines the advantages of neural networks and HMM using a multiple hypothesis (or N-best) paradigm. The connectionist component of the system, the segmental neural net (SNN), models all the frames of a phonetic segment simultaneously, thus overcoming the well-known conditional-independence limitation of the HMM. They describe the hybrid system and discuss various aspects of SNN modeling, including network architectures, training algorithms and context modeling. Finally, they evaluate the hybrid system by performing several speaker-independent experiments with the DARPA Resource Management (RM) corpus, and demonstrate that the hybrid system shows a consistent improvement in performance over the baseline HMM system.