Search

Cognitive agents are typically utilized in autonomous systems for automated decision making. With the widespread use of autonomous systems in complex environments, the need for real-time cognitive agents is essential. Cognitive agents are more capable when they are able to process larger amounts of information to make more informed and intelligent decisions. The solution search space for cognitive agents increases exponentially with large volumes of varied data. In this paper, we present the parallelization of the knowledge-mining component of a cognitive agent that can be programmed to reason like humans. This study examined a novel high-performance path-based forward checking algorithm on 128 compute nodes at the Ohio Supercomputing Center (768 cores) to achieve a speedup of over 200 times compared to a serial implementation of our algorithm. The serial implementation is around 10–25 times faster than a conventional Java-based constraint solver at generating the first solution.

Declarative memory enables cognitive agents to effectively store and retrieve factual memory in real-time. Increasing the capacity of a real-time agent's declarative memory increases an agent's ability to interact intelligently with its environment but requires a scalable retrieval system. This work represents an extension of the Accelerated Declarative Memory (ADM) system, referred to as Hardware Accelerated Declarative Memory (HADM), to execute retrievals on a GPU. HADM also presents improvements over ADM's CPU execution and considers critical behavior for indefinitely running declarative memories. The negative effects of a constant maximum associative strength are considered, and mitigating solutions are proposed. HADM utilizes a GPU to process the entire semantic network in parallel during retrievals, yielding significantly faster declarative retrievals. The resulting GPU-accelerated retrievals show an average speedup of approximately 70 times over the previous Service Oriented Architecture Declarative Memory (soaDM) implementation and an average speedup of approximately 5 times over ADM. HADM is the first GPU-accelerated declarative memory system in existence.

In many cases, low power autonomous systems need to make decisions extremely efficiently. However, as a problem space becomes more complex, finding a solution quickly becomes nearly impossible using traditional computing methods. Thus, in this work we show that constraint satisfaction problems (CSPs) can be solved quickly and efficiently using spiking neural networks. Constraint satisfaction is a general problem solving technique that can be applied to a large number of different applications. To demonstrate the validity of this algorithm, we show successful execution of the Boolean satisfiability problem (SAT) on the Intel Loihi spiking neuromorphic research processor. In many cases, constraint satisfaction problems have solution sets as opposed to single solutions. Therefore, the manycore architecture of the Loihi chip is used to parallelize the solution finding process, leading to a quasi-complete solution set generated at extreme efficiency (dynamic energy as low as 8 micro joules per solution). Power consumption in this spiking processor is due primarily to the propagation of spikes, which are the key drivers of data movement and processing. Thus, the proposed SAT algorithm was customized for spiking neural networks to achieve the greatest efficiency gains. To the best of our knowledge, the work in this paper exhibits the first implementation of constraint satisfaction on a low power embedded neuromorphic processor capable of generating a solution set. In general, we show that embedded spiking neuromorphic hardware is capable parallelizing the constraint satisfaction problem solving process to yield extreme gains in terms of time, power, and energy.

Cognitive agents are typically utilized in autonomous systems for automated decision making. These systems interact at real time with their environment and are generally heavily power constrained. Thus, there is a strong need for a real time agent running on a low power platform. The agent examined is the Cognitively Enhanced Complex Event Processing (CECEP) architecture. This is an autonomous decision support tool that reasons like humans and enables enhanced agent-based decision-making. It has applications in a large variety of domains including autonomous systems, operations research, intelligence analysis, and data mining. One of the most time consuming and key components of CECEP is the mining of knowledge from a repository described as a Cognitive Domain Ontology (CDO). One problem that is often tasked to CDOs is asset allocation. Given the number of possible solutions in this allocation problem, determining the optimal solution via CDO can be very time and energy consuming. A grid of isolated spiking neurons is capable of generating solutions to this problem very quickly, although some degree approximation is required to achieve the speedup. The approximate spiking approach presented in this work was able to complete nearly all allocation simulations with greater than 98% accuracy. Our results in this work show that constraining the possible solution space by creating specific rules for a scenario can alter the quality of the allocation result. We present a study compares allocation score and computation time for three different constraint implementation cases. Given the vast increase in speed, as well as the reduction computational requirements, the presented algorithm is ideal for moving asset allocation to low power embedded hardware.

Asset allocation is a compute intensive combinatorial optimization problem commonly tasked to autonomous decision making systems. However, cognitive agents interact in real time with their environment and are generally heavily power constrained. Thus, there is strong need for a real time asset allocation agent running on a low power computing platform to ensure efficiency and portability. As an alternative to traditional techniques, work presented in this paper describes how spiking neuron algorithms can be used to carry out asset allocation. We show that a significant reduction in computation time can be gained if the user is willing to accept a near optimal solution using our spiking neuron approach. As of late, specialized neuromorphic spiking processors have demonstrated a dramatic reduction in power consumption relative to traditional processing techniques for certain applications. Improved efficiencies are primarily due to unique algorithmic processing that produces a reduction in data movement and an increase in parallel computation. In this work, we use the TrueNorth spiking neural network processor to implement our asset allocation algorithm. With an operating power of approximately 50 mW, we show the feasibility of performing portable low-power task allocation on a spiking neuromorphic processor.

Cognitive agents are typically utilized in autonomous systems for automated decision making. These systems interact at real time with their environment and are generally heavily power constrained. Thus, there is a strong need for a real time agent running on a low power platform. The agent examined is the Cognitively Enhanced Complex Event Processing (CECEP) architecture. This is an autonomous decision support tool that reasons like humans and enables enhanced agent-based decision-making. It has applications in a large variety of domains including autonomous systems, operations research, intelligence analysis, and data mining. One of the key components of CECEP is the mining of knowledge from a repository described as a Cognitive Domain Ontology (CDO). One problem that is often tasked to CDOs is asset allocation. Given the number of possible solutions in this allocation problem, determining the optimal solution via CDO can be very time consuming. In this work we show that a grid of isolated spiking neurons is capable of generating solutions to this problem very quickly, although some degree of approximation is required to achieve the speedup. However, the approximate spiking approach presented in this work was able to complete all allocation simulations with greater than 99.9% accuracy. To show the feasibility of low power implementation, this algorithm was executed using the Intel Loihi manycore neuromorphic processor. Given the vast increase in speed (greater than 1000 times in larger allocation problems), as well as the reduction in computational requirements, the presented algorithm is ideal for moving asset allocation to low power, portable, embedded hardware., Comment: Accepted and to appear in the proceedings of the 2019 IEEE-INNS International Joint Conference on Neural Networks. Citation: C. Yakopcic, T. Atahary, N. Rahman, T. M. Taha, A. Beigh, and S. Douglass, High Speed Approximate Cognitive Domain Ontologies for Asset Allocation Using Loihi Spiking Neurons, IEEE-INNS International Joint Conference on Neural Networks, Budapest, Hungary, July, 2019

The cognitively enhanced complex event processing (CECEP) architecture being developed at the US Air Force is an autonomous decision support tool that reasons and learns like humans and enables enhanced agent-based decision making. It has applications in both military and civilian domains. One of the most computationally challenging aspects of CECEP is mining domain knowledge captured in cognitive domain ontologies (CDOs). Real-time agents require massively linked knowledge databases to be searched using a large set of constraints to generate intelligent decisions in run time. This study examines how to parallelize and accelerate the CDO knowledge mining algorithm by converting it into a constraint network that can be solved using a parallelized generate and test exhaustive depth first search algorithm. Our results show that 1 GPGPU can provide speed-ups of 100 times over a Xeon CPU core and almost 8 times over a Xeon Phi processor for the search algorithm.

Cognitive agents are typically utilized in autonomous systems for automated decision making. These systems interact in real time with their environment and are generally heavily power constrained. Thus, there is a strong need for a real time agent running on a low power computing platform. An autonomous agent reasons like humans and enables enhanced agent-based decision-making. The agent examined is the Cognitively Enhanced Complex Event Processing (CECEP) architecture. Task allocation is one of most time consuming and power-hungry processes within the CECEP architecture. Task allocation is a combinatorial optimization problem in the field of operations research. In the CECEP architecture, the cognitive agent performs this task allocation. This paper examines the performance of two task allocation algorithms (the greedy algorithm and the auction algorithm) that have the potential to significantly reduce the power and energy required to perform this task. Thus, it may be possible to execute complex task allocation problems on custom, portable, low power devices.

Cognitive agents are typically utilized in autonomous systems for automated decision making. These systems interact at real time with their environment and are generally heavily power constrained. Thus, there is a strong need for a real time agent running on a low power platform. The agent examined is the Cognitively Enhanced Complex Event Processing (CECEP) architecture. This is an autonomous decision support tool that reasons like humans and enables enhanced agent-based decision-making. It has applications in a large variety of domains including autonomous systems, operations research, intelligence analysis, and data mining. One of the most time consuming and key components of CECEP is the mining of knowledge from a repository described as a Cognitive Domain Ontology (CDO). One problem that is often tasked to CDOs is asset allocation. Given the number of possible solutions in this allocation problem, determining the optimal solution via CDO can be very time consuming. In this work we show that a grid of isolated spiking neurons is capable of generating solutions to this problem very quickly, although some degree approximation is required to achieve the speedup. However, the approximate spiking approach presented in this work was able to complete all allocation simulations with greater than 98% accuracy. Given the vast increase in speed (greater than 1000 times in some cases), as well as the reduction computational requirements, the presented algorithm is ideal for moving asset allocation to low power embedded hardware.

Cognitive computing can enable automatic decision making, thus enabling the design of more intelligent and autonomous systems. An autonomous cognitive decision support architecture, Cognitively Enhanced Complex Event Processing (CECEP), was developed at United States Air Force Research Laboratory to enable agent-based decision-making. CECEP has several components, which allow cognitive agents to make automatic decision without any human interaction. One of the key components, Cognitive Domain Ontologies (CDO) are a knowledge repository for CECEP agents. The knowledge mining process from CDOs in the CECEP architecture is analogous to solving a constraint-satisfaction problem (CSP). Cognitive agents are more capable when they are able to process or mine larger volume of information to make more informed and intelligent decisions. This means that the CDOs need to process large knowledge bases. Cognitive agents need to consider environmental inputs and make decisions at runtime. Hence CDOs with large knowledge bases need to be accelerated for real time cognitive agents. This study introduces an accelerated CDO solver with multiple optimizations for generating and ranking solutions for an autonomous cognitive agent. As CDOs typically generate all equally weighted valid solutions, it is impossible to find the best one from this solution set without imposing some relative importance of certain criteria. This study implemented several optimization conditions through objective functions and processed them as Multiple Constraint Optimization Problems (MCOPs) to rank solutions on a GPGPU. It achieved a speedup of 85–100 times over existing solvers for ranking the solutions.

Cognitive agents are typically utilized in autonomous systems for automated decision making. These systems interact in real time with their environments and are generally heavily power constrained. Thus there is a strong need for a real time agent running on a low power computing platform for these systems. This paper examines how some of the components of a cognitive agent can be mapped onto the IBM TrueNorth neurosynaptic system to achieve real time performance at low power. The agent examined is the Cognitively Enhanced Complex Event Processing (CECEP) architecture. Although it is geared towards autonomous decision making, CECEP also has applications in operations research, intelligence analysis, and data mining. One of the key components of CECEP is the Cognitive Domain Ontology (CDO), used for mining decisions from a large knowledge repository based on situational inputs and constraints. As CDOs are the most time and power consuming part of CECEP, we implemented them on the TrueNorth processor by mapping the solution space of CDOs into a pattern matching form.

Cognitive agents are typically utilized in autonomous systems for automated decision making. These systems interact at real time with their environment and are generally heavily power constrained. Thus, there is a strong need for a real time agent running on a low power platform. This paper examines how some of the components of a cognitive agent can be executed in memristor crossbar circuits capable of highly parallel instruction execution. The agent examined is the Cognitively Enhanced Complex Event Processing (CECEP) architecture. This is an autonomous decision support tool that reasons like humans and enables enhanced agent-based decision-making. It has applications in a large variety of domains including autonomous systems, operations research, intelligence analysis, and data mining. One of the most time consuming and key components of CECEP is the mining of knowledge from a repository described as a Cognitive Domain Ontology (CDO). We show that CDOs can be implemented using memristor crossbars using two different approaches. The first is a lookup table approach that is stored in a high density memristor matching circuit. The second in a multilayer perceptron implementation that uses an ex-situ memristor based neuromorphic system. In each case, the example CDOs are implemented successfully.

Cognitive agents are typically utilized in autonomous systems for automated decision making. These systems interact at real time with their environment and are generally heavily power constrained. Thus there is a strong need for a real time agent running on a low power computing platform. This paper examines how some of the components of a cognitive agent can be mapped onto the IBM TrueNorth neurosynaptic system to achieve real time performance at low power. The agent examined is the Cognitively Enhanced Complex Event Processing (CECEP) architecture. This is an autonomous decision support tool that reasons like humans and enables enhanced agent-based decision-making. It has applications in a large variety of domains including autonomous systems, operations research, intelligence analysis, and data mining. One of the most time consuming and key components of CECEP is the mining of knowledge from a repository described as a Cognitive Domain Ontology (CDO). This paper looks at how CDOs can be mapped into neural form for realtime applications, and then examines their implementation on the IBM TrueNorth processor. The results show that the TrueNorth implementation generated correct answers for all the CDOs examined.

Cognitive agents are typically utilized in autonomous systems for automated decision making. These systems interact at real time with their environment and are generally heavily power constrained. Thus, there is a strong need for a real time agent running on a low power platform. This paper examines how some of the components of a cognitive agent can be mapped onto lookup tables stored in memristor crossbar circuits capable of highly parallel string matching. The agent examined is the Cognitively Enhanced Complex Event Processing (CECEP) architecture. This is an autonomous decision support tool that reasons like humans and enables enhanced agent-based decision-making. It has applications in a large variety of domains including autonomous systems, operations research, intelligence analysis, and data mining. One of the most time consuming and key components of CECEP is the mining of knowledge from a repository described as a Cognitive Domain Ontology (CDO). We show that CDOs can be implemented in lookup tables, and present a memristor crossbar circuit capable of storing these lookup tables very efficiently.

This paper describes the acceleration of the declarative knowledge retrieval system of a cognitive architecture, namely ACT-R. The core of ACT-R's retrieval mechanism, activation calculation, is accelerated through leveraging the speed of C++ and the MapReduce program model. Work described in this paper represents an extension of previous Erlang-based concurrent activation. ACT-R's retrieval process is re-examined and optimized in this solution. Concurrency available in the execution platform is exploited to maximize the acceleration of declarative retrieval. The resulting implementation, referred to as Accelerated Declarative Memory (ADM), presents a highperformance activation calculation that enables practical use of more massive declarative memories. ADM presents new mechanisms to access and traverse declarative memory to reduce the overhead of executing retrievals. This solution offers retrieval latencies 20 times faster than the previous Erlang solution.

The Cognitively Enhanced Complex Event Processing (CECEP) architecture is an autonomous decision support tool that reasons and learns like humans and enables enhanced agent-based decision making. This architecture has applications in both military and civilian domains, including operations research and data mining. One of the most computationally challenging aspects of CECEP is mining domain knowledge captured in cognitive domain ontologies (CDOs). This requires massively linked knowledge databases to be searched based on a large set of constraints to generate intelligent decisions. The solution search for CDOs is a kind of Constraint Satisfaction Problem (CSP). Due to the unique nature of the CDO-to-CSP mapping, generic constraint solvers are inefficient. We introduce a novel high performance path based forward checking CSP algorithm to solve CDOs and compare it to a commonly utilized CSP solving application. Additionally our approach enables a very compact and efficient representation of the entire solution space to allow fast processing and give agents key ideas on how to further reduce the solution space. The proposed algorithm provided a speedup of about 10–25 times for generating the first solution and about half a million times for all the solutions over a Choco based conventional CSP approach.

Agent-based decision aids can improve their performance by mining domain knowledge captured in cognitive domain ontologies (CDOs). The CDOs can be specialized to represent knowledge from any of a large variety of domains, including network security. Search algorithms mine data from these CDOs using constraints. While complex CDOs increase the capabilities of decision agents, they are computationally expensive to mine. Complex CDOs can function as powerful decision aids if they can be processed in realistic time frames. To achieve this, it is necessary to parallelize and accelerate the CDO knowledge mining algorithms. This chapter introduces CDOs and examines how they can be transformed into constraint networks for processing on high performance compute platforms. The constraint networks were solved using a parallelized generate and test exhaustive depth first search algorithm. Two compute platforms for acceleration are examined: Intel Xeon multicore processors, and NVIDIA graphics processors (GPGPUs). This chapter shows that 1 NVIDIA Tesla C2070 GPGPU can provide a speed-up of 100 times over 1 Xeon CPU core for the search algorithm. The scaling of the algorithm on a high performance GPGPU cluster achieved estimated speed-ups of over 1,000 times.

Functional Magnetic Resonance Imaging(fMRI) is a noninvasive imaging technique that measures the hemodynamic response related to neural activity in the human brain. Event-related functional magnetic resonance imaging (efMRI) is a form of functional Magnetic Resonance Imaging (fMRI) in which a series of fMRI images are time-locked to a stimulus presentation and averaged together over many trials. Again an event related potential (ERP) is a measured brain response that is directly the result of a thought or perception. Here the neuronal response of human visual cortex in normal healthy patients have been studied. The patients were asked to perform a visual three choice reaction task; from the relative response of each patient corresponding neuronal activity in visual cortex was imaged. The average number of neurons in the adult human primary visual cortex, in each hemisphere has been estimated at around 140 million. Statistical analysis of this experiment was done with SPM5(Statistical Parametric Mapping version 5) software. The result shows a robust design of imaging the neuronal activity of human visual cortex., {"references":["Rosen BR, Buckner RL, Dale AM (1998). \"Event-related functional\nMRI: past, present, and future.\" Proc Natl Acad Sci USA 95:773-780.","Marc A. Burock, Randy L. Buckner, Marty G. Woldorff, Bruce R.\nRosen, Anders M. Dale (19 November,1998). \"Randomized eventrelated\nexperimental designs allow for extremely rapid presentation rates\nusing functional MRI.\" Brain Imaging Neuroreport: 3735-3739.","Visual Cortex, Available: http://en.wikipedia.org/wiki/Visual_cortex","Participants, Available:\nhttp://www.fmridc.org/f/fmridc/protege/exp_protocols.html?uid=912265","Visual Single Task: Available:\nhttp://www.fmridc.org/f/fmridc/protege/exp_protocols.html?uid=912269","Scanning Procedure: Available\nhttp://www.fmridc.org/f/fmridc/protege/scanner_protocols.html?uid=47\n6463","Scanning Procedure: Available:\nhttp://www.fmridc.org/f/fmridc/protege/scanner_protocols.html?uid=47\n6464","Preprocessing:Available http://www.fil.ion.ucl.ac.uk/spm","SPM5Manual:Available\nhttp://www.fil.ion.ucl.ac.uk/spm/doc/manual.pdf\n[10] Rafael C. Gonzalez, Richard E. Woods (2002). Digital Image\nProcessing. Pearson Education Inc: 110-120.\n[11] Petter Jezzard, Paul M.Mathews, Stephen M. Smith(2001). Functional\nMRI-An Introduction to Methods. Oxford University Press: 180-186.\n[12] R. Henson, W. Penny(2005). \"ANOVAs and SPM.\"\n[13] F-Test: Avaiable : http://en.wikipedia.org/wiki/F-test\n[14] Petter Jezzard, Paul M.Mathews, Stephen M. Smith(2001). Functional\nMRI-An Introduction to Methods. Oxford University Press: 220-222"]}

Mobile IP, the current method of internet connectivity is most often found in WLAN environments where users need to carry their mobile devices across multiple LANs with different IP address. This project work first surveys existing protocols for supporting IP mobility and then proposes an extension to mobile IP architecture, called Robust Hierarchical Mobile IP version6 (RH-MIPv6). This architecture attempts to achieve smaller handoff latency in intra-domain mobile network. In RH-MIPv6 a mobile node (MN) registers primary (P-RCoA) and secondary (S-RCoA) regional care of address to two different MAPs simultaneously. In adaptation to this a “MOVING AREA BASED MAP SELECTION SCHEME” is proposed in this paper. A mechanism is developed to enable the mobile node or correspondent node to detect the failure of primary MAP and change their attachment from primary to secondary MAP. With this recovery procedure, it is possible to reduce the failure recovery time. In this paper it is shown that RH-MIPv6 has faster recovery time than MIPv4 and HMIPv6.