In this thesis, we present a methodology for designing computationally efficient algorithms for large-scale inference systems on system architectures with distributed autonomous agents. The principle of information-based computation is the underlying idea driving elements of this methodology. The methodology consists of a layered information processing architecture, which organizes specific inference tasks into layers so that the information extracted from these tasks can be exploited to design an efficient inference system, and the hierarchy, which is a representation combining elements of circuits and branching programs from which efficient distributed local algorithms implementing global tasks on general system architectures can be generated. The general issues addressed by this methodology are: (1) how to efficiently collect distributed information for estimation when collecting information is expensive, (2) how to distribute a global algorithm among multiple agents, and (3) how to take advantage of prior computations to aid in subsequent computations. To demonstrate that the methodology is a good one, we present an application of sensor networks to battlefield awareness. The problem is to gain intelligence about targets in the environment from the information collected by a set of sensor nodes which are distributed over a field and have real physical constraints on energy and communication. In particular...
If we are to understand human-level intelligence, we need to understand how meanings can be learned without explicit instruction. I take a step toward that understanding by focusing on the symbol-grounding problem, showing how symbols can emerge from a system that looks for regularity in the experiences of its visual and proprioceptive sensory systems. More specifically, my implemented system builds descriptions up from low-level perceptual information and, without supervision, discovers regularities in that information. Then, my system, with supervision, associates the regularity with symbolic tags. Experiments conducted with the implementation shows that it successfully learns symbols corresponding to blocks in a simple 2D blocks world, and learns to associate the position of its eye with the position of its arm. In the course of this work, I take a new perspective on how to design knowledge representations, one that grapples with the internal semantics of systems, and I propose a model of an adaptive knowledge representation scheme that is intrinsic to the model and not parasitic on meanings captured in some external system, such as the head of a human investigator.; by Stephen David Larson.; Thesis (M. Eng.)--Massachusetts Institute of Technology...
If we are ever to have intelligent systems, they will need memory. Memory is the core of learning; intelligence is about entering, extracting, and synthesizing its contents. What makes the memory problem difficult is that memory is not a simple collection of facts. The how and why of where those facts were acquired is a key part of how they are internalized and used later. As a step towards solving this large and difficult problem, I have focused on how people learn to trust each other when they have a conversation. The model I have created represents people as sets of self-organizing maps; each has a map to represent his own beliefs, and a map to represent what he thinks of another person. Beliefs are in this model restricted to likes and dislikes, across a wide range of topics. In this thesis I describe the program implemented in Java to test this model. The model has been tested on four different kinds of conversations. with the topics of animals and cars, to determine whether its behavior looks reasonable to a human observer. In this work I show how a simple, natural model can closely approximate human behavior. without need for tweaking parameters.; by Catherine Miller.; Thesis (M. Eng.)--Massachusetts Institute of Technology...
One common characteristic of all intelligent life is continuous perceptual input. A decade ago, simply recording and storing a a few minutes of full frame-rate NTSC video required special hardware. Today, an inexpensive personal computer can process video in real-time tracking and recording information about multiple objects for extended periods of time, which fundamentally enables this research. This thesis is about Perceptual Data Mining (PDM), the primary goal of which is to create a real-time, autonomous perception system that can be introduced into a wide variety of environments and, through experience, learn to model the activity in that environment. The PDM framework infers as much as possible about the presence, type, identity, location, appearance, and activity of each active object in an environment from multiple video sources, without explicit supervision. PDM is a bottom-up, data-driven approach that is built on a novel, robust attention mechanism that reliably detects moving objects in a wide variety of environments. A correspondence system tracks objects through time and across multiple sensors producing sets of observations of objects that correspond to the same object in extended environments. Using a co-occurrence modeling technique that exploits the variation exhibited by objects as they move through the environment...
Human intelligence is a product of cooperation among many different specialists. Much of this cooperation must be learned, but we do not yet have a mechanism that explains how this might happen for the "high-level" agile cooperation that permeates our daily lives. I propose that the various specialists learn to cooperate by learning to communicate, basing this proposal on the phenomenon of communication bootstrapping, in which shared experiences form a basis for agreement on a system of signals. In this dissertation, I lay out a roadmap for investigating this hypothesis, identifying problems that must be overcome in order to understand the capabilities of communication bootstrapping and in order to test whether it is exploited by human intelligence. I then demonstrate progress along the course of investigation laid out in my roadmap: * I establish a measure of developmental cost that allows me to eliminate many possible designs * I develop a method of engineering devices for use in models of intelligence, including characterizing their behavior under a wide variety of conditions and compensating for their misbehavior using failure simplification. * I develop mechanisms that reliably produce communication bootstrapping such that it can be used to connect specialists in an engineered system. * I construct a demonstration system including a simulated world and pair of observers that learn world dynamics via communication bootstrapping.; by Jacob Stuart Michael Beal.; Thesis (Ph. D.)--Massachusetts Institute of Technology...
If we are to understand human-level cognition, we must understand how the mind finds the patterns that underlie the incomplete, noisy, and ambiguous data from our senses and that allow us to generalize our experiences to new situations. A wide variety of commercial applications face similar issues: industries from health services to business intelligence to oil field exploration critically depend on their ability to find patterns in vast amounts of data and use those patterns to make accurate predictions. Probabilistic inference provides a unified, systematic framework for specifying and solving these problems. Recent work has demonstrated the great value of probabilistic models defined over complex, structured domains. However, our ability to imagine probabilistic models has far outstripped our ability to programmatically manipulate them and to effectively implement inference, limiting the complexity of the problems that we can solve in practice. This thesis presents BLAISE, a novel framework for composable probabilistic modeling and inference, designed to address these limitations. BLAISE has three components: * The BLAISE State-Density-Kernel (SDK) graphical modeling language that generalizes factor graphs by: (1) explicitly representing inference algorithms (and their locality) using a new type of graph node...
Social media is becoming increasingly important in society and culture, empowering consumers to group together on common interests and share opinions through the Internet. The social web shifts the originators of content from companies to users. Differences caused by this dynamic result in existing web analytic techniques being inadequate. Because people reveal their thoughts and preferences in social media, there are significant opportunities in business intelligence by analyzing social media. These opportunities include brand monitoring; trend recognition, and targeted advertising. The market for social media analytics in business intelligence is further validated by its direct application in the consumer research market. Challenges lie ahead for development and adoption of social media analytics. Technology used in these analytics, such as natural language processing and social network analysis, need to mature to improve accuracy, performance, and scalability. Nevertheless, social media continues to grow at a rapid pace, and organizations should form strategies to incorporate social media analytics into their business intelligence frameworks.; by Bobby Lo.; Thesis (M. Eng.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science...
Approved for public release; distribution unlimited.; Developing large hard-real-time systems in a traditional way usually creates inconsistencies among the user's needs, the requirements and the implementation. Rapid Prototyping by using Prototype System Description Language (PSDL) and Computer Aided Prototyping System (CAPS) minimizes the time and resource costs, and maximizes reliability. In this technique, designer builds the prototype with the initial requirements, and the user evaluates the actual behavior of the prototype against its expected behavior. If prototype fails to execute properly, the user and the designer work together to change the requirements and the prototype, unit the prototype captures the critical aspects of the software system. This thesis uses the rapid prototyping approach to produce an Ada software prototype of C3I workstation, which provides commonality and connectivity between naval platforms and land bases by providing the ability to process tactical data from many interfaces in real-time. The major emphasis of the prototype is to support C3I information management functions, message generation and information display.; Ltjg., Turkish Navy; Ltjg., Turkish Navy
Approved for public release; distribution is unlimited; This research entails the design and development of an automated system that allows researchers working remotely to store, manage, and transfer assertion data to an external system run by the Intelligence Advanced Research Projects Activity. The research stems from the University of Maryland's involvement in the Social-Cultural Content in Language program, which seeks to investigate methodologies, designs, and technologies that can contribute in the understanding of the social goals of persons or groups of people by demonstrating a relationship between these goals and their particular language use. In this research, we interview the stakeholders to determine the software requirements of the system. After a careful analysis of the requirements, we used the Unified Modeling Language notation to provide the reader a visual model of the software design. Finally, we develop a working prototype of the proposed system consisting of two Web services and a Web service client written in the Java programming language.; US Marine Corps (USMC) author
Approved for public release; distribution is unlimited; The area of concern addressed by this research is the development of an Automated Tactical Operations Command. Control, Communications, and Intelligence Planning Tool (ATOC3IPT) to aid commanders and their staffs in the decision-making process. The tool is based on Hyper-NPSNET, an application which integrates a single level hypermedia overlay into a 3D virtual world with at most a single instance of each type of multimedia information available at each 3D location. However, Tactical Operations Centers (TOCOs) require multiple overlays, each with possibly multiple instances of each type of multimedia information available at each location. Also Hyper-NPSNET is a single user system whereas the TOCs require a multiple user system with restricted access. Tactical information display and query retrieval requirement of the command and control organizations were studied. A database structure and control hierarchy, were designed. An appropriate graphical user interface (GUI) was developed. The hypersystem used in Hyper-NPSNET was extended to include multiple, permission-protected overlays with multiple instances of each type of multimedia information available at each location. The resulting ATOC3IPT is a battlefield planning tool which incorporates today's technology. New hypermedia information display...
Approved for release; distribution is unlimited; Computer network operations (CNO) can be considered a relatively new phenomenon being encount modern warfare. Computer network operation is comprised of three components, computer network attack computer network exploitation (CNE), and computer network defense (CND). Computer network attack is def operations to disrupt, deny, degrade, or destroy information resident in computer networks, or the computers and ne themselves. Computer network exploitation is the intelligence collection and enabling operations to gather data from adversary automated information systems (AIS) or networks. Finally, computer network defense are those me internal to the protected entity, taken to protect and defend information, computers, and networks from disruption, degradation, or destruction. No longer is warfare limited to the use of kinetic weapons and conventional methods of war. Computer network operations have become an integral part of our adversary's arsenal and more attention must be paid to the effects of CNO activities, particularly CNA and CNE being conducted by our adversaries. Of the many states suspected of conducting active CNO activities against the United States and other nations, none warrants more attention than North Korea. This thesis presents the development of methodology using information available from open sources. This work is intended to prove that a useful methodology for assessing the CNO capabilities and limitations of North Korea can be developed using only open source information.; Lieutenant...
Cognitive science is founded on the conjecture that natural intelligence can be explained in terms of computation. Yet, notoriously, there is no consensus among philosophers of cognitive science as to how computation should be characterised. While there are subtle differences between the various accounts of computation found in the literature, the largest fracture exists between those that unpack computation in semantic terms (and hence view computation as the processing of representations) and those, such as that defended by Chalmers (2011), that cleave towards a purely syntactic formulation (and hence view computation in terms of abstract functional organisation). It will be the main contention of this paper that this dispute arises because contemporary computer science is an amalgam of two different historical traditions, each of which has developed its own proprietary conception of computation. Once these historical trajectories have been properly delineated, and the motivations behind the associated conceptions of computation revealed, it becomes a little clearer which should form the foundation for cognitive science.; http://j-cs.org/issues/__vol012i4/3.html; Gerard O’Brien
As robots begin to emerge from the cloisters of industrial and military applications and enter the realms of coöperative partners for people, one of the most important facets of human-robot interaction (HRI) will be communication. This can not merely be summarized in terms of the ongoing development into unimodal communication mechanisms such as speech interfaces, which can apply to any technology. Robots will be able to communicate in physically copresent, "faceto-face" interactions across more concurrent modalities than any previous technology. Like many other technologies, these robots will change the way people work and live, yet we must strive to adapt robots to humans, rather than the reverse. This thesis therefore contributes mechanisms for facilitating and influencing human-robot communication, with an explicit focus on the most salient aspect that differentiates robots from other technologies: their bodies. In order to communicate effectively with humans, robots require supportive infrastructure beyond the communications capabilities themselves, much as do the humans themselves. They need to be able to achieve basic common ground with their counterparts in order to ensure that accurate and efficient communication can occur at all.; (cont.) For certain types of higher level communication...
A multi-mode software system contains several distinct modes of operation and a controller for deciding when to switch between modes. Even when developers rigorously test a multi-mode system before deployment, they cannot foresee and test for every possible usage scenario. As a result, unexpected situations in which the program fails or underperforms (for example, by choosing a non-optimal mode) may arise. This research aims to mitigate such problems by training programs to select more appropriate modes during new situations. The technique, called program steering, creates a new mode selector by learning and extrapolating from previously successful experiences. Such a strategy, which generalizes the knowledge that a programmer has built into the system, may select an appropriate mode even when the original programmer had not considered the scenario. We applied the technique on simulated fish programs from MIT's Embodied Intelligence class and on robot control programs written in a month-long programming competition. The experiments show that the technique is domain independent and that augmenting programs via program steering can have a substantial positive effect on their performance in new environments.; by Lee Chuan Lin.; Thesis (M. Eng.)--Massachusetts Institute of Technology...
This thesis investigates the opportunity of teaching people how to cook by analyzing the ingredients' chemical content as they are using them, and the consequent creation of a specific class of context-aware cookware that aids its users. An inquisition on the chemical content of different food and the appropriate electronics for measuring it was done. An instrument, with embedded sensors and intelligence and in the form of a spatula, was created base on the result of the research, and tested to be able to measure salinity, acidity, temperature, and consistency. This tool was used to demonstrate that several ingredients could be measured easily, and recipes as varied as pickles and pancakes could be improved. The work demonstrates the possibility of having intelligence in the kitchen, and examines the pedagogical value of intelligent tools when they are capable of collaborating with and guiding its user. The research also inquires into the field of ubiquitous computing, in which sensors are placed in ordinary objects, and to assess its impact in a domestic environment.; by Mansim Connie Cheng.; Thesis (M. Eng.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 2003.; Includes bibliographical references (p. 81-82).
Evidence-based medicine employs expert opinion and clinical data to inform clinical decision making. The objective of this study is to determine whether it is possible to complement these sources of evidence with information about physician “group intelligence” that exists in electronic health records. Specifically, we measured laboratory test “repeat intervals”, defined as the amount of time it takes for a physician to repeat a test that was previously ordered for the same patient. Our assumption is that while the result of a test is a direct measure of one marker of a patient's health, the physician's decision to order the test is based on multiple factors including past experience, available treatment options, and information about the patient that might not be coded in the electronic health record. By examining repeat intervals in aggregate over large numbers of patients, we show that it is possible to 1) determine what laboratory test results physicians consider “normal”, 2) identify subpopulations of patients that deviate from the norm, and 3) identify situations where laboratory tests are over-ordered. We used laboratory tests as just one example of how physician group intelligence can be used to support evidence based medicine in a way that is automated and continually updated.
In recent years Business Intelligence (BI) systems have consistently been
rated as one of the highest priorities of Information Systems (IS) and business
leaders. BI allows firms to apply information for supporting their processes
and decisions by combining its capabilities in both of organizational and
technical issues. Many of companies are being spent a significant portion of
its IT budgets on business intelligence and related technology. Evaluation of
BI readiness is vital because it serves two important goals. First, it shows
gaps areas where company is not ready to proceed with its BI efforts. By
identifying BI readiness gaps, we can avoid wasting time and resources. Second,
the evaluation guides us what we need to close the gaps and implement BI with a
high probability of success. This paper proposes to present an overview of BI
and necessities for evaluation of readiness. Key words: Business intelligence,
Evaluation, Success, Readiness; Comment: International Journal of Computer Science & Engineering Survey
(IJCSES) Vol.3, No.2, April 2012
In this paper, we continue the efforts of the Computational Theory of
Intelligence (CTI) by extending concepts to include computational processes in
terms of Genetic Algorithms (GA's) and Turing Machines (TM's). Active, Passive,
and Hybrid Computational Intelligence processes are also introduced and
discussed. We consider the ramifications of the assumptions of CTI with regard
to the qualities of reproduction and virility. Applications to Biology,
Computer Science and Cyber Security are also discussed.; Comment: Total of 5 figures. This paper was originally presented at RAMSA 2013
in Visakhaptnam, India in December 2013
Collective Intelligence (CI) is the ability of a group to exhibit greater
intelligence than its individual members. Expressed by the common saying that
"two minds are better than one," CI has been a topic of interest for social
psychology and the information sciences. Computer mediation adds a new element
in the form of distributed networks and group support systems. These facilitate
highly organized group activities that were all but impossible before computer
mediation. This paper presents experimental findings on group problem solving
where a distributed software system automatically integrates input from many
humans. In order to quantify Collective Intelligence, we compare the
performance of groups to individuals when solving a mathematically formalized
problem. This study shows that groups can outperform individuals on difficult
but not easy problems, though groups are slower to produce solutions. The
subjects are 57 university students. The task is the 8-Puzzle sliding tile
game.; Comment: University of California, Santa Cruz Tech Report UCSC-CRL-02-28
nformation security is an issue of global concern. As the Internet is
delivering great convenience and benefits to the modern society, the rapidly
increasing connectivity and accessibility to the Internet is also posing a
serious threat to security and privacy, to individuals, organizations, and
nations alike. Finding effective ways to detect, prevent, and respond to
intrusions and hacker attacks of networked computers and information systems.
This paper presents a knowledge discovery frame work to detect DoS attacks at
the boundary controllers (routers). The idea is to use machine learning
approach to discover network features that can depict the state of the network
connection. Using important network data (DoS relevant features), we have
developed kernel machine based and soft computing detection mechanisms that
achieve high detection accuracies. We also present our work of identifying DoS
pertinent features and evaluating the applicability of these features in
detecting novel DoS attacks. Architecture for detecting DoS attacks at the
router is presented. We demonstrate that highly efficient and accurate
signature based classifiers can be constructed by using important network
features and machine learning techniques to detect DoS attacks at the boundary
controllers.; Comment: IEEE Publication format...