Abstract: The utilization of electronic medical record (EMR) data to establish the disease diagnosis model has become an important research content of biomedical informatics. Deep learning can automatically extract features from the massive data, which brings about breakthroughs in the study of EMR data. The challenge is that deep learning lacks semantic knowledge, which leads to impracticability in medical science. This research proposes a method of incorporating lexical-semantic knowledge from abundant entities into a convolutional neural network (CNN) framework for pediatric disease diagnosis. Firstly, medical terms are vectorized into Lexical Semantic Vectors (LSV), which are concatenated with the embedded word vectors of word2vec to enrich the feature representation. Secondly, the semantic distribution of medical terms serves as Semantic Decision Guide (SDG) for the optimization of deep learning models. The study evaluates the performance of LSV-SDG-CNN model on four kinds of Chinese EMR datasets. Additionally, CNN, LSV-CNN, and SDG-CNN are designed as baseline models for comparison. The experimental results show that LSV-SDG-CNN model outperforms baseline models on four kinds of Chinese EMR datasets. The best configuration of the model yielded an F1 score of 86.20%. The results clearly demonstrate that CNN has been effectively guided and optimized by lexical-semantic knowledge, and LSV-SDG-CNN model improves the disease classification accuracy with a clear margin.
Abstract: This paper studies a case where the targeted surface roughness of fused deposition modeling (FDM) additive manufacturing process is improved. The process is designing to reduce or eliminate the defects and improve the process capability index Cp and Cpk for an FDM additive manufacturing process. The baseline Cp is 0.274 and Cpk is 0.654. This research utilizes the Taguchi methodology, to eliminate defects and improve the process. The Taguchi method is used to optimize the additive manufacturing process and printing parameters that affect the targeted surface roughness of FDM additive manufacturing. The Taguchi L9 orthogonal array is used to organize the parameters' (four controllable parameters and one non-controllable parameter) effectiveness on the FDM additive manufacturing process. The four controllable parameters are nozzle temperature [°C], layer thickness [mm], nozzle speed [mm/s], and extruder speed [%]. The non-controllable parameter is the environmental temperature [°C]. After the optimization of the parameters, a confirmation print was printed to prove that the results can reduce the amount of defects and improve the process capability index Cp from 0.274 to 1.605 and the Cpk from 0.654 to 1.233 for the FDM additive manufacturing process. The final results confirmed that the Taguchi methodology is sufficient to improve the surface roughness of FDM additive manufacturing process.
Abstract: Since polymerase chain reaction (PCR) has been
invented, it has emerged as a powerful tool in genetic analysis. The
PCR products are closely linked with thermal cycles. Therefore, to
reduce the reaction time and make temperature distribution uniform in
the reaction chamber, a novel oscillatory thermal cycler is designed.
The sample is placed in a fixed chamber, and three constant isothermal
zones are established and lined in the system. The sample is oscillated
and contacted with three different isothermal zones to complete
thermal cycles. This study presents the design of the geometric
characteristics of the chamber. The commercial software
CFD-ACE+TM is utilized to investigate the influences of various
materials, heating times, chamber volumes, and moving speed of the
chamber on the temperature distributions inside the chamber. The
chamber moves at a specific velocity and the boundary conditions
with time variations are related to the moving speed. Whereas the
chamber moves, the boundary is specified at the conditions of the
convection or the uniform temperature. The user subroutines compiled
by the FORTRAN language are used to make the numerical results
realistically. Results show that the reaction chamber with a rectangular
prism is heated on six faces; the effects of various moving speeds of
the chamber on the temperature distributions are examined. Regarding
to the temperature profiles and the standard deviation of the
temperature at the Y-cut cross section, the non-uniform temperature
inside chamber is found as the moving speed is larger than 0.01 m/s.
By reducing the heating faces to four, the standard deviation of the
temperature of the reaction chamber is under 1.4×10-3K with the range
of velocities between 0.0001 m/s and 1 m/s. The nature convective
boundary conditions are set at all boundaries while the chamber moves
between two heaters, the effects of various moving velocities of the
chamber on the temperature distributions are negligible at the assigned
time duration.
Abstract: The competitive learning is an adaptive process in
which the neurons in a neural network gradually become sensitive to
different input pattern clusters. The basic idea behind the Kohonen-s
Self-Organizing Feature Maps (SOFM) is competitive learning.
SOFM can generate mappings from high-dimensional signal spaces
to lower dimensional topological structures. The main features of this
kind of mappings are topology preserving, feature mappings and
probability distribution approximation of input patterns. To overcome
some limitations of SOFM, e.g., a fixed number of neural units and a
topology of fixed dimensionality, Growing Self-Organizing Neural
Network (GSONN) can be used. GSONN can change its topological
structure during learning. It grows by learning and shrinks by
forgetting. To speed up the training and convergence, a new variant
of GSONN, twin growing cell structures (TGCS) is presented here.
This paper first gives an introduction to competitive learning, SOFM
and its variants. Then, we discuss some GSONN with fixed
dimensionality, which include growing cell structures, its variants
and the author-s model: TGCS. It is ended with some testing results
comparison and conclusions.
Abstract: State-dependent Riccati equation based controllers are
becoming increasingly popular because of having attractive
properties like optimality, stability and robustness. This paper focuses
on the design of a roll autopilot for a fin stabilized and canard
controlled 122mm artillery rocket using state-dependent Riccati
equation technique. Initial spin is imparted to rocket during launch
and it quickly decays due to straight tail fins. After the spin phase, the
roll orientation of rocket is brought to zero with the canard deflection
commands generated by the roll autopilot. Roll autopilot has been
developed by considering uncoupled roll, pitch and yaw channels.
The canard actuator is modeled as a second-order nonlinear system.
Elements of the state weighing matrix for Riccati equation have been
chosen to be state dependent to exploit the design flexibility offered
by the Riccati equation technique. Simulation results under varying
conditions of flight demonstrate the wide operating range of the
proposed autopilot.