Browsing by Subject "Neural Networks"

Now showing 1 - 7 of 7

Open Access
Identifying probability distributions using neural networks
(1995) Yılmaz, Anıl
Economics deal with real life phenomena by constructing representative models o f the system being questioned. Input data provide the driving force for such models. The requirement o f identifying the underlying distributions of data sets is encountered in economics on numerous occasions. Most of the time, after the collection o f the raw data, the underlying statistical distribution is sought by the aid o f nonparametric statistical methods. At this step o f the problem, the feasibility of using neural networks for identification o f probability distributions is investigated. Also, for this purpose, a comparison with the traditional goodness o f fit tests is carried out in this study.
Open Access
Learning translation templates for closely related languages
(Springer, Berlin, Heidelberg, 2003) Altıntaş, Kemal; Güvenir, H. Altay
Many researchers have worked on example-based machine translation and different techniques have been investigated in the area. In literature, a method of using translation templates learned from bilingual example pairs was proposed. The paper investigates the possibility of applying the same idea for close languages where word order is preserved. In addition to applying the original algorithm for example pairs, we believe that the similarities between the translated sentences may always be learned as atomic translations. Since the word order is almost always preserved, there is no need to have any previous knowledge to identify the corresponding differences. The paper concludes that applying this method for close languages may improve the performance of the system.
Open Access
Multiplication free neural networks
(2018-01) Mallah, Maen M. A.
Artificial Neural Networks, commonly known as Neural Networks (NNs), have become popular in the last decade for their achievable accuracies due to their ability to generalize and respond to unexpected patterns. In general, NNs are computationally expensive. This thesis presents the implementation of a class of NN that do not require multiplication operations. We describe an implementation of a Multiplication Free Neural Network (MFNN), in which multiplication operations are replaced by additions and sign operations. This thesis focuses on the FPGA and ASIC implementation of the MFNN using VHDL. A detailed description of the proposed hardware design of both NNs and MFNNs is analyzed. We compare 3 dfferent hardware designs of the neuron (serial, parallel and hybrid), based on latency/hardware resources trade-off. We show that one-hidden-layer MFNNs achieve the same accuracy as its counterpart NN using the same number of neurons. The hardware implementation shows that MFNNs are more energy efficient than the ordinary NNs, because multiplication is more computationally demanding compared to addition and sign operations. MFNNs save a significant amount of energy without degrading the accuracy. The fixed-point quantization is discussed along with the number of bits required for both NNs and MFNNs to achieve floating-point recognition performance.
Open Access
Parallel mapping and circuit partitioning heuristics based on mean field annealing
(1992) Bultan, Tevfik
Moan Field Annealinp; (MFA) aJgoritlim, receñí,ly proposc'd for solving com binatorial optimization problems, combines the characteristics of nenral networks and simulated annealing. In this thesis, MFA is formulated for tlie mapping i)roblcm and the circuit partitioning problem. EHicient implementation schemes, which decrease the complexity of the proposed algorithms by asymptotical factors, are also given. Perlormances of the proposed MFA algorithms are evaluated in comparison with two well-known heuristics: simulated annealing and Kernighan-Lin. Results of the experiments indicate that MFA can be used as an alternative heuristic for the mapping problem and the circuit partitioning problem. Inherent parallelism of the MFA is exploited by designing efficient parallel algorithms for the proposed MFA heuristics. Parallel MFA algorithms proposed for solving the circuit partitioning problem are implemented on an iPS(J/2’ hypercube multicompute.r. Experimental results show that the proposed heuristics can be efficiently parallelized, which is crucial for algorithms that solve such computationally hard problems.
Open Access
Prediction with expert advice: on the role of contexts, bandit feedback and risk-awareness
(2018-12) Ekşioğlu, Kubilay
Along with the rapid growth in the size of data generated and collected over time, the need for developing online algorithms that can provide answers without any offline training has considerably increased. In this thesis, we consider the prediction with expert advice problem under the online learning framework. Specifically, we consider problems where experts have asymmetric information about the sample space. First, we propose an algorithm that selects a subset of the experts and makes predictions based on the advices of this subset. Then, we propose another algorithm that clusters samples in an online manner and makes predictions based on the history of observations and decisions within each cluster. Next, we consider the Safe Bandit, a variant of the Risk Aware Multi Armed Bandit, where the goal is to minimize the number of rounds in which a risky arm is chosen. Adopting mean-variance as the risk notion, we define an arm as risky if its mean-variance is higher than a given threshold. Using this, we define a new regret measure called Risk Violation Regret (RVR), which depends on the number of times risky arms are selected. Then, we propose a learning algorithm called Exploration and Exploitation with Risk Thresholds (EXERT), and prove that it achieves O(1) RVR with high probability. Afterwards, we use EXERT in an expert selection problem, where each expert corresponds to a neural network with reject option. For this, we propose a method to train these neural networks and use them to evaluate the performance of EXERT in real-world datasets.
Open Access
Scheduling with artificial neural networks
(1993) Gürgün, Burçkaan
Artificial Neural Networks (ANNs) attempt to emulate the massively parallel and distributed processing of the human brain. They are being examined for a variety of problems that have been very difficult to solve. The objective of this thesis is to review the current applications of ANNs to scheduling problems and to develop a parallelized network model for solving the single machine mean tardiness scheduling problem and the problem of finding the minimum makespan in a job-shop. The proposed model is also compared with the existing heuristic procedures under a variety of experimental conditions.
Open Access
Synthesis of artificial neural networks by transconductors only
(Kluwer Academic Publishers, 1991) Tan, M.A.
Hardware implementation of artificial neural networks has been attracting great attention recently. In this work, the analog VLSI implementation of artificial neural networks by using only transconductors is presented. The signal flow graph approach is used in synthesis. The neural flow graph is defined. Synthesis of various neural network configurations by means of neural flow graph is described. The approach presented in this work is technology independent. This approach can be applied to new neural network topologies to be proposed or used with transconductors designed in future technologies. © 1991 Kluwer Academic Publishers.