W. Kenneth Jenkins

Transfer entropy between cortical and basal ganglia electrophysiology

Linear measures such as cross-correlation, coherence, and directed transfer functions have previously been applied to investigate the functional connectivity between brain regions. However, such methods do not account for nonlinear interactions between the signals. Separately, dopaminergic cell transplants have been shown to provide symptomatic amelioration and partial electrophysiological normalization of aberrant basal ganglia firing patterns in Parkinsons Disease. However, the precise extent and mechanisms of basal ganglia electrophysiological normalization have remained unclear. In this experiment we computed the transfer entropy between electroencephalograms (EEGs) and basal ganglia local field potentials (LFPs) from urethane-anesthetized rats, in order to investigate both linear and nonlinear interactions. We used the 6-hydroxy-dopamine lesioned medial forebrain bundle hemiparkinsonian (HP) rat model, and recorded from the substantia nigra and subthalamic nucleus of normal rats, HP rats, and HP rats with murine fetal ventral mesencephalic cell transplants, looking separately at slow wave EEG epochs versus global activation epochs. We found that both the crosscorrelation and the transfer entropy between the motor cortical EEG and basal ganglia LFPs was increased in the HP group (p<;0.05) and returned to normal levels in the grafted group, in most nuclei and conditions. However, the transfer entropy more robustly showed the difference between the groups. Our findings indicate that transfer entropy is a sensitive tool for nonlinear inter-nucleic functional connectivity analyses, and demonstrate the novel restorative ability of dopaminergic grafts for the parkinsonian basal ganglia electrophysiology.

Today's Engineering Education Is a Liberal Arts Education of the Future [Point of View]

Many times when an electrical engineer is visiting with nontechnical members of the extended family, some who are senior citizens and some who are not, the engineer is frequently told ‘‘all electrical devices that I use on a daily basis (laptop computers, iPads, Chromebooks, Kindles, wireless printers, cloud printers, fax machines, scanners, iPhones, etc.) are so frustrating that I am ready to throw them out the window!’’ The next question the engineer often hears is ‘‘Since you are an electrical engineer can you fix my wireless printer connections and get it working?’’ The engineer’s typical response is ‘‘I may not know all the details about your particular printer, but I can explain why it may or may not be working.’’ Occasionally, if the engineer has two hours to read through documentation downloaded from the Internet and works with the printer settings, he or she may get the device working. But more often than not the engineer’s final words of advice are ‘‘I recommend that you call the Geek Squad to get this problem fixed’’ [1]. The typical response from the nontechnical family member is ‘‘Oh, I didn’t realize I would have to get professional help to get my printer connected.’’ It is now time for us to think about how society should help our average citizens overcome such frustrating daily experiences that come with the advanced technologies that are invading our daily lives. Let’s now back up in history and consider the question as to ‘‘What was the purpose of a liberal arts education in the past and why did it become a central stream within our society?’’ This first point to be considered here is that a liberal arts education does not enable a person to practice a particular profession such as medicine, law, engineering, accounting, etc. Each of these professions requires special training that is needed to practice and become certified in the field. Although a liberal arts education does not prepare students to practice specific professions, it does provide many educational skills that enable people to function well and have a comfortable life within our society [2]. The first educational skill provided by a liberal arts education is the art of reading, writing, and speaking. These are communication skills that are needed to efficiently interact with many types of citizens on a diversified basis. Good communication skills are needed to successfully engage in professional activities, social interactions, religious discussions, and pressing family affairs. A second educational skill involves acquiring knowledge of history to understand and appreciate where society came from and to accurately predict where it is going. A third educational skill is a basic knowledge and appreciation of fine arts (music, painting, sculpture, Digital Object Identifier: 10.1109/JPROC.2014.2341311

Facial recognition with PCA and machine learning methods

Facial recognition is a challenging problem in image processing and machine learning areas. Since widespread applications of facial recognition make it a valuable research topic, this work tries to develop some new facial recognition systems that have both high recognition accuracy and fast running speed. Efforts are made to design facial recognition systems by combining different algorithms. Comparisons and evaluations of recognition accuracy and running speed show that PCA + SVM achieves the best recognition result, which is over 95% for certain training data and eigenface sizes. Also, PCA + KNN achieves the balance between recognition accuracy and running speed.

Multiscale Autoregressive Identification of Neuroelectrophysiological Systems

Electrical signals between connected neural nuclei are difficult to model because of the complexity and high number of paths within the brain. Simple parametric models are therefore often used. A multiscale version of the autoregressive with exogenous input (MS-ARX) model has recently been developed which allows selection of the optimal amount of filtering and decimation depending on the signal-to-noise ratio and degree of predictability. In this paper, we apply the MS-ARX model to cortical electroencephalograms and subthalamic local field potentials simultaneously recorded from anesthetized rodent brains. We demonstrate that the MS-ARX model produces better predictions than traditional ARX modeling. We also adapt the MS-ARX results to show differences in internuclei predictability between normal rats and rats with 6OHDA-induced parkinsonism, indicating that this method may have broad applicability to other neuroelectrophysiological studies.

Generalized Data-Aided Equalizers

The generalized data-aided equalizer (GDAE) is a hybrid structure that combines attractive features of the adaptive canceller-equalizer (ACE) and the decision-feedback equalizer (DFE) structures. With limited hardware complexity, generally speaking, the performance of a data-aided equalizer depends on the channel impulse response. Because the structure of an ACE or DFE is fixed, it is not feasible to reconfigure either structure according to the channel response. However, with the GDAE and an appropriate optimization scheme, it is possible to adjust the structure to obtain the best performance. The basic idea of the GDAE is to apply an additional adaptive system to select the best filter structure for a given channel characteristics. This is possible by using a feedback filter bank. This filter bank decouples the decision feedback so that the feedback can eliminate any ISI components.