- 科学新闻
- 特写新闻
- 小型脑器官 " 阅读 " 盲文标志着生物线性计算的关键步骤
小型脑器官 " 读取 " 盲文标志着生物线性计算的关键步骤
一组科学家已经表明，脑有机物可以传递与盲文字母对应的信号，这是朝向开发未来节能生物碱化计算机迈出的重要一步。 大脑有机物与记录不同盲文字母的触觉信号的传感器远程连接，产生每个字母不同的信号，使一个解码器能够区分给有机物的不同字母。 虽然单个脑器官提供了准确的可解码信号 61%的时间，但一个三人组成的团队表现明显更好，产生了准确的83%的时间信号。 三人小组在区分信号与人工噪声方面也比the decoder在没有有机体参与的情况下直接研究模糊信号更好。
电脑能包括小脑器官吗？人类大脑是一台令人难以置信的节能计算机，硅处理器试图模仿它实现巨大的节能，而不是传统的处理器。 现在科学家正在尝试使用以人类前脑为模型的有机体来开发节能生物碱计算。 朝着这个方向迈出的关键的第一步是了解有机体能否可靠地接收和转发信号，一个以布里斯托尔为基地的研究人员小组利用盲文字母对此进行了调查。 他们发现由三个脑器官组成的团队产生的信号准确对应了83%的时间的不同字母。
称 Tianyi Liu是布里斯托尔大学博士候选人，也是"人工智能前沿"文章的第一作者。 “这为我们提供了将人造传感器与生物神经系统连接起来的实用方法，而生物神经系统是通向更节能计算的一个非常有趣的可能途径。”
从A到盲文
刘和她的团队测试的Biohybrid系统使用放置在多电极阵列上的大脑有机体，并包含在一个能保持有机体生命并监测其反应的系统中。 这些有机体横跨不到一毫米，缺乏人类大脑拥有的几种结构，但含有约10,000个神经元，以及辅助细胞。 多电极阵列允许它们接收电气信号。
首先，科学家对盲文字母表的26个字母应用了一种记录触觉数据的传感器。 然后他们把这些数据转换成电刺激模式，并呈现出不同模式的有机体，看看它们是否会产生神经活性，可以告诉科学家他们得到了什么刺激。 换句话说，他们是否能够 " 阅读 " 与每个盲文字母对应的电讯信号，并产生一个与他们得到的信息相对应的信号？
有机体的神经活动模式是使用简单的机器学习算法，即辅助向量机进行分类的，它可以对规律进行排序，并检测给定的信号是否与特定字母对应。 单个有机体始终生成一个信号，可以准确解码为大约61%的时间的字母。 然而，三个有机体一起工作，产生了大约83%的时间可以识别的信号。
“经有机体处理后，触觉数据仍然包含足够的信息，足以准确分类盲文字母，”Bristol大学的Benjamin Ward-Cherrier博士是文章的资深作者。 “若干有机物的组合产出也显示性能有所改善。”
阅读和下载原文章
三个脑袋比一个好？
研究者还不能说明为什么三个有机体的表现优于一个；这可能是因为有机体之间的个别差异意味着它们相辅相成。
“不同的有机物可能具有略微不同的网络连接、发育状态和持续活动，因此同样的输入可以产生类似但并非相同的反应，” “把这些答复合并，可能给我们提供我们无法从一个有机体中获得的补充信息。 同时，如果每个有机体的行为都不同，显然更难建立一个可预测和标准化的系统。”
为了使任务更加艰巨，科学家们对盲文字母信号采用了不同的噪声过滤器。 他们还使用同样的辅助矢量机来比较器官响应的解码与故意模糊信号的直接解码。 机器在没有有机体参与的情况下直接解码，结果比从有机体反应中解码更容易被愚弄——它不太能够识别人工噪声中的字母信号。
关闭循环
科学家们强调，这些实验并没有显示有机体能否随时间而学习；它们只是表明有机体收到并响应盲文信号。
“我们表明，触觉信息可以编码、通过有机物处理和解码，但我们没有证明有机物的学习或长期变化。” 称 Liu。 “我们还用固定编码办法和相对简单的神经活动量度测试了一项触觉任务。 下一步肯定要超越盲文；我们感兴趣的是纹理，边缘，以及各种触摸，并最终在闭路任务中，器官从它的工作中得到反馈。 这将使我们能够开始问这个有机体是否能够通过经验实际适应。”
(原始内容存档于2019-09-21). REPUBLING Guidelines：开放接入和共享研究是Frontiers任务的一部分。 除非另有说明，你可以重新转载在"前沿报"新闻网站发布的文章——只要你包含回原研究的链接。 不得出售文章。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
- Science news
- Featured news
- Mini brain organoids ‘reading’ Braille marks key step towards biohybrid computing
Mini brain organoids ‘reading’ Braille marks key step towards biohybrid computing
A group of scientists have shown that brain organoids can pass on signals that correspond to Braille letters, an essential step towards developing the energy-efficient biohybrid computers of the future. The brain organoids, remotely connected to a sensor that recorded tactile signals from different Braille letters, produced signals that were different for each letter, allowing a decoder to distinguish between the different letters the organoids had been given. Although individual brain organoids provided accurately decodable signals 61% of the time, a team of three performed significantly better, producing an accurate signal 83% of the time. The team of three was also better at distinguishing the signals from artificial noise than the decoder working directly on the obscured signals, without the organoids’ involvement.
Could computers include tiny brain organoids one day? The human brain is an incredibly energy-efficient computer, and silicon processors that try to imitate it achieve huge energy savings over traditional processors. Now scientists are experimenting with using organoids modelled on the human forebrain to develop energy-efficient biohybrid computing. A crucial first step towards this is learning whether the organoids can receive and relay signals reliably, which a Bristol-based team of researchers investigated using Braille letters. They found that teams of three brain organoids produced signals which accurately corresponded to different letters 83% of the time.
“We can take information from a real tactile sensor, turn it into a form that a living neural network can receive, and then recover meaningful information from the organoid response,” said Tianyi Liu, PhD candidate at Bristol University and first author of the article in Frontiers in Artificial Intelligence. “That gives us a practical way to connect artificial sensors with biological neural systems, which are a very interesting possible route towards more energy-efficient computing.”
From A to Braille
The biohybrid systems tested by Liu and her team use brain organoids placed on multielectrode arrays and contained in a system that keeps the organoids alive and monitors their responses. These organoids are less than a millimeter across and lack several structures that a human brain has, but contain about 10,000 neurons, as well as supporting cells. The multielectrode arrays allow them to receive electrical signals.
First, the scientists applied a sensor that records tactile data to the 26 letters of the Braille alphabet. Then they converted this data to electrical stimulation patterns and presented the organoids with the different patterns to see if they would produce neural activity that could tell the scientists what stimulus they’d received. In other words, were they able to ‘read’ the electrical signal that corresponded to each Braille letter, and produce a signal that recognizably corresponded to the information they’d been given?
The organoids’ patterns of neural activity were classified using a simple machine learning algorithm, a support vector machine, that could sort the patterns and detect whether given signals corresponded to specific letters. A single organoid consistently produced a signal that could be accurately decoded as a letter about 61% of the time. However, three organoids working together produced a recognizable signal about 83% of the time.
“After processing by the organoids, the tactile data still contains enough information to classify Braille letters with good accuracy,” said Dr Benjamin Ward-Cherrier of Bristol University, senior author of the article. “Combining outputs from several organoids also showed an improvement in performance.”
Read and download the original article
Three brains are better than one?
The researchers can’t yet tell why three organoids performed better than just one; it could be because individual variation between the organoids means that they complement each other.
“Different organoids may have slightly different network connections, developmental states, and ongoing activity, so the same input can produce similar but not identical responses,” said Liu. “Combining those responses may give us complementary information that we would not get from a single organoid. At the same time, if every organoid behaves differently, it obviously makes it harder to build a predictable and standardized system.”
To make the task harder, the scientists applied different noise filters to the Braille letter signals. They also used the same support vector machine to compare decoding from the organoid responses with direct decoding of the deliberately obscured signals. The machine decoding directly, without the organoids’ involvement, turned out to be easier to fool than decoding from the organoid responses — it was less able to identify the letter signals among the artificial noise.
Closing the loop
The scientists stress that these experiments don’t show if organoids can learn over time; they just show that the organoids received and responded to the Braille signals.
“We showed that tactile information could be encoded, processed through the organoids, and decoded, but we did not demonstrate learning or long-term changes in the organoids,” said Liu. “We also tested one tactile task using a fixed encoding scheme and a relatively simple measure of neural activity. The next step is definitely to move beyond Braille; we are interested in things such as textures, edges, and different kinds of touch, and eventually in closed-loop tasks where the organoid receives feedback from what it does. That would allow us to start asking whether the organoid can actually adapt through experience.”
REPUBLISHING GUIDELINES: Open access and sharing research is part of Frontiers’ mission. Unless otherwise noted, you can republish articles posted in the Frontiers news site — as long as you include a link back to the original research. Selling the articles is not allowed.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.