Page 9 - IJB-6-1

Yu and Jiang
A B

Figure 1. Typical three-dimensional printing process (a) and extrusion-based bioprinting methods (b).

force (piston-driven dispensing), or rotation in the algorithm will be improved and then
(screw-driven dispensing). generate a machine learning model. Using the
Machine learning is an emerging technology updated machine learning parameters, it can then
that can optimize systems through smarter and predict the results with new input data. The most
effective use of products, materials, and services. commonly used machine learning methods include
[33]
[34]
In terms of 3D printing processes, machine supervised learning , unsupervised learning ,
[35]
learning can lead to a reduction of fabrication time, and reinforcement learning .
minimized cost, and increased quality. In literature, In supervised learning, the training data are
machine learning has already been applied to a collection of x, y form pairs, and the objective
ʌ
process optimization [17-21] , dimensional accuracy is to get the predicted result y in response to a
analysis [22-25] , manufacturing defect detection [26-28] , query x. x, y can be more than one element that
and material property prediction [29-32] . However, will be expressed as a vector in machine learning.
machine learning has not been applied in 3D Currently, supervised learning has been used in
bioprinting yet. In this paper, the perspective on spam classification of email, medical diagnosis
how machine learning can help to improve 3D systems for patients and face recognition over
bioprinting is discussed. Related machine learning images.
used in 3D printing will be briefly reviewed for In unsupervised learning, the input data are
illustrating its effects on 3D bioprinting. We unlabeled data which are different from supervised
believe that machine learning can significantly learning. Algorithms will automatically learn
affect the future development of 3D bioprinting and extract the features of the input data and
and hope this paper can inspire some ideas on then divide them into different clusters. The
how machine learning can be used to improve 3D aim of unsupervised learning is to model the
bioprinting. underlying distribution or structure of the input
data for learning more about the data. Currently,
2 Machine learning unsupervised learning has been applied in market
segmentation for targeting appropriate customers,
Machine learning is one of today’s most rapidly clustering documents based on content, image
growing technical fields. It is a subset of artificial division, and anomaly or fraud detection in
intelligence, mainly focusing on the designing banking companies.
of systems. Machine learning allows these In reinforcement learning, the information from
designed systems to learn and make predictions the training data fed into algorithms is intermediate
based on the previous experience which is data between unsupervised and supervised learning.
in terms of machines. Figure 2 shows a typical Instead of indicating the correct output for a given
machine learning process. The data in the training input in supervised learning, the training data are
set needs to be trained first by the algorithm. assumed to provide only an indication as to whether
During the training process, the parameters an action is correct or not. Currently, reinforcement

International Journal of Bioprinting (2020)–Volume 6, Issue 1 5