Difference between revisions of "Team:SEU/Improve"

 
(One intermediate revision by the same user not shown)
Line 71: Line 71:
 
                                       <div>
 
                                       <div>
 
                                           <h1>Improve</h1>
 
                                           <h1>Improve</h1>
−
                                           <p>This is our first time to participate in iGEM. According to our survey, there is few similar projects in previous iGEM competition hence our focus in this page lies in improving state-of-the-art molecular computing neural networks in literature [1].</p>
+
                                           <p>This is our first time to participate in iGEM. According to our survey, there is few similar projects in previous iGEM competition hence our focus in this project lies in improving state-of-the-art molecular neural networks in literature [1].</p>
−
                                           <p>In the previous papar, inputs to neural networks are binary i.e. the values are either 0 or 1, which limits the application. In computer science applications, data are usually real numbers. If we are able to support real value calculation, CPU/GPU-based software applications can be possibly mapped. Also, more importantly, training is not integrated to the system. It is one critical part of neural networks, and totally relying on <i>in silico</i> training cannot fully exploit the potential of molecular computing. Hence we make such improvements:</p>
+
                                           <p>In [1], the values of input signals of neural networks are binary i.e. the values are constrained to be either 0 or 1, which limits the application in real world. In computer science applications, data are usually real numbers. In addition, training, which is one critical part of artificial intelligence systems, has not been integrated to the molecular system in [1]. Therefore we make such improvements:</p>
−
                                           <p>1. Employ continuous values during computation.</p>
+
                                           <p>1. We support continuous value computation in our molecular systems. CPU/GPU-based software applications can be possibly mapped to our system based on this feature. There is a pattern recognition example in <a href="https://2019.igem.org/Team:SEU/Demonstrate"></a>Demonstrate Page.</a> </p>
−
                                           <p>2. Integrate training to our neuron.</p>
+
                                           <p>2. We integrate training to our neural networks. Therefore our system does not require in silico training processes.</p>
 
                                           <p>Also, we develop a <a href="https://2019.igem.org/Team:SEU/Software"><b>Software Tool</b></a>to help other researchers design their own systems.</p>
 
                                           <p>Also, we develop a <a href="https://2019.igem.org/Team:SEU/Software"><b>Software Tool</b></a>to help other researchers design their own systems.</p>
 
                                           <h3>Reference</h3>
 
                                           <h3>Reference</h3>

Latest revision as of 13:45, 18 October 2019





Improve

This is our first time to participate in iGEM. According to our survey, there is few similar projects in previous iGEM competition hence our focus in this project lies in improving state-of-the-art molecular neural networks in literature [1].

In [1], the values of input signals of neural networks are binary i.e. the values are constrained to be either 0 or 1, which limits the application in real world. In computer science applications, data are usually real numbers. In addition, training, which is one critical part of artificial intelligence systems, has not been integrated to the molecular system in [1]. Therefore we make such improvements:

1. We support continuous value computation in our molecular systems. CPU/GPU-based software applications can be possibly mapped to our system based on this feature. There is a pattern recognition example in Demonstrate Page.

2. We integrate training to our neural networks. Therefore our system does not require in silico training processes.

Also, we develop a Software Toolto help other researchers design their own systems.

Reference

[1] K. Cherry, L. Qian, "Scaling up molecular pattern recognition with DNA-based winner-take-all neural networks," Nature, vol. 559, no. 7714, pp.370-376, 2018.