深度学习中的深度是指什么_什么是深度学习
深度学习中的深度是指什么
Data science is transforming various domains such as robotics medicine and everything in between. This transformation is largely attributed to advancements in research computational capabilities professional interests within the domain and the expanding toolkit of data science techniques. Typically individuals perceive data science as representing significant advancements within artificial intelligence (AI) ultimately envisioning robots achieving tasks traditionally dominated by humans independently.
数据科学正以前所未有的方式重塑多个领域。 涵盖从机器人技术延伸至医学的整个领域范围。 这场变革在一定程度上得益于研究活动、计算能力的提升、对该领域的持续兴趣以及日益完善的工具包。 人们普遍认为数据科学是人工智能发展过程中的一个显著里程碑。 这种趋势最终实现了机器人执行那些传统上由人类主导的任务。
Although this could be an aspect of data science, it does not encompass all aspects of the field. Instead, AI serves as a component within the broader domain of data science. Various fields including machine learning (ML) and artificial intelligence have evolved into prominent areas becoming integral components within the growing landscape of modern analytics.
虽然说这可能构成数据科学的一个方面内容,但其实它并非数据科学的全部内容。反过来说的话,人工智能确实是数据科学工具箱中的一员,而机器学习（ML）以及人工智能等领域如今已成为数据科学研究的重点方向
These tools are strongly effective because they acquire and evolve to improve the results in response to tasks which they are assigned. This is crucial because although humans are capable of learning and adapting to optimize outcomes, machines currently hold an advantage in scaling up these processes.
这些工具之所以强大,是因为它们能够学习并适应以实现对所负责情况的优化. 这一现象的重要性体现在,尽管人类同样可以学习并适应以优化结果,但机器在更为广泛的范围内展现出这一优势.
Many problems are quite complexed, making it unreasonable for humans to attempt solving them individually. Instead, individuals should utilize their understanding of a situation alongside their computational resources and data to derive meaningful solutions. Rather than relying on human effort alone, individuals should make full use of their situational knowledge combined with computational capabilities and vast amounts of data to achieve significant outcomes. At the convergence point lies Deep Learning, a powerful tool that revolutionizes how we approach complex challenges.
许多问题极为复杂，在这方面难以合理地解决这些问题。相反地，在这一点上是深度学习！它不仅能够有效地处理这些问题本身，并且还能够充分利用它们对于具体情况的理解以及计算能力与数据的结合来实现预期结果。
Image by Trist’n Joseph 图片由Trist'n Joseph
Deep learning represents a specialized technique within artificial intelligence. As a subset of machine learning, it focuses on enabling systems to learn from data and improve through experience. Artificial intelligence refers to any method that allows computers to emulate human behavior. The term suggests it’s a specialized field within computer science focused on replicating intelligent processes in machines. While this approach has shown remarkable success across various industries, understanding its underlying principles and objectives is essential before delving into deep learning. In this context, intelligence refers not only to the ability but also encompasses the capability of processing information that aids in making informed decisions.
那么什么是深度学习呢？从广义上说，则它就是人工智能技术的一种表现形式。既然深度学习属于人工智能领域的一部分，则我们首先得弄清楚人工智能的概念及其目标所在。人工智能是指能够模拟人类行为以完成特定任务的技术。顾名思义的话，则它属于计算机科学的一个分支领域，并且其核心目标就是使计算机能够像人类一样具备智能特质。再次强调这一点很重要吗？或者是否应该仅仅关注其内部智能的发展呢？在这种情况下，则可以把智能看作处理信息以支持未来决策的能力。这些信息则可用于为未来的决策提供依据
Hence, the objective of artificial intelligence (AI) is to develop effective algorithms capable of handling information that aids in making informed decisions. Machine learning (ML) is commonly employed to achieve this objective. AI includes machine learning as one of its subsets, enabling systems to be designed for autonomous enhancement of performance based on experiences accumulated through data analysis.
为了实现支持未来决策制定的有效信息处理能力的开发目的，人工智能旨在设计出高效的算法。机器学习（ML）通常被用于通过机器学习技术实现这一目标。作为人工智能的一个重要分支领域，在不需要显式编程的情况下使系统能够自主学习并持续改进基于先前经验的能力。
Now the concept of 'learning' has garnered considerable attention in recent discussions. However, unlike how humans engage in reading books, conducting research or posing thoughtful inquiries, machines learn differently from humans by processing vast amounts of data through algorithms instead of relying on predefined models. Rather, machine learning algorithms employ computational techniques to extract meaningful insights directly from raw data without assuming predefined relationships between variables.
如今"学习"已成为一个热门话题。然而，在模仿人类的学习方面目前仍面临诸多挑战。相比之下，在模仿人类的学习方面目前仍面临诸多挑战。相反，在深度学习框架下机器可以直接分析海量数据并提取其本质特征。
To accomplish this, the algorithms are designed to identify patterns within data and create a target function that most effectively maps an input variable x to a target variable y. Deep learning represents a subset that builds upon this concept. The objective of deep learning is for systems to automatically extract the most crucial pieces of information necessary for making informed future decisions.
为此, 我们必须运用算法识别数据中的规律并构建目标函数, 该目标函数应将输入变量_x_映射至输出变量_y_. 深度学习可视为这种理念的一个延伸领域, 其主要目的在于自动生成最有效的信息提取机制, 为其未来的决策提供理论依据
Image by Trist’n Joseph 图片由Trist'n Joseph
The concept of advancing machine learning further might seem abstract, potentially blurring the distinction between machine learning and deep learning. This suggests that traditional machine learning approaches focus on establishing rule sets within datasets through manual engineering. As such, these methods may fall short of expectations when deployed outside controlled development environments.
进一步推进这一抽象的想法可能会导致对机器学习与深度学习之间区别的模糊理解；这使得机器学习与深度学习之间的区别变得不那么明显；这是一个典型的机器学习算法试图从数据中提取一组通用规则的过程；而这些通用规则通常是人工设计而非自然产生的；因此，在开发环境中应用时效果可能达不到预期的效果水平
Consider ‘ ThisIsPizza ’, which is my fictional app to detect whether the object in a picture is a pizza slice or not. Now, having an app and an algorithm which can accurately determine whether an object is pizza is important because one would not want to eat a triangular object which looks like pizza but is not pizza. Pizza is very complexed, and recall that machines are better at dealing with complexed situations than humans are. Therefore, the classification rule could then be if there is a triangular object with at least tomato sauce, cheese, pepperoni, and a crust at the base of the triangle, then it is a pizza. But then, the obvious question would be, how to determine whether something is tomato sauce, cheese, and pepperoni?
此应用程序用于检测图片中的对象是否为披萨片。“ThisIsPizza”是我的虚构应用程序名称。准确确定对象是否为比萨对人们来说至关重要——他们显然不会想要看起来像披萨但实际上却是三角形的对象。考虑到这一点非常重要：如果检测到一个三角形物体，并且该物体含有番茄酱、奶酪、意大利辣香肠以及三角形底面外壳，则被认为是披萨。然而，在现实中如何确定番茄酱、奶酪和意大利辣香肠的存在是一个显而易见的问题？
Deep learning aims to learn these features solely from raw data. Pepponi is not defined as a red, circular image. The question arises: can we define cheese or tomato sauce? The deep learning model constructs a hierarchical representation of lines, curvature, and other features for distinguishing between cheese and tomato sauce. These lower-level features are combined to detect higher-level constructs like pizza slices.
深度学习的核心是主要从原始数据中提取这些功能。 不需要将意大利辣香肠视为淡红色圆形图像。 您是否能够预见到对奶酪或番茄酱进行定义？ 而实际在开发线、曲率等特征时会建立层次表示形式；这些特征可用于区分奶酪和番茄酱；随后通过组合这些特征来识别比萨饼切片。
Image by Trist’n Joseph 图片由Trist'n Joseph
Despite ThisIsPizza being employed as a simple illustration, the concepts it elucidates are extensively applied in real-world scenarios to address multifaceted challenges. Grasping the issues tackled by ThisIsPizza, particularly how deep learning can mitigate these issues, forms an integral part of comprehending the essence of deep learning.
虽然 ThisIsPizza 被当作玩具示例，在日常应用中仍采用了涉及的概念来克服多个现实世界中的难题。 掌握 ThisIsPizza 面临的难题以及通过深入学习的方法有效地应对这些难题对于理解什么是深度学习至关重要。
Let us examine a more specific instance: Apple’s Face ID system. This innovative technology is designed to identify and verify users’ faces for unlocking devices and securing services like Apple Pay. By avoiding unnecessary complexity, we can focus on the core functionality of facial recognition. The system deploys advanced cameras to capture facial features accurately, but the real question is: How does it recognize a face? To put it another way, imagine if your smartphone mistakenly scanned your leg instead of your face; payment transactions could be compromised.
为了更具体地理解这一概念，请参考苹果的Face ID技术。这是一个先进的面部识别系统（ facial recognition system），它不仅能够帮助用户快速解锁设备（ unlock devices），还能在支付等服务中起到身份验证的作用（ serve as an authentication system in services like Apple Pay）。在不深入讨论细节的情况下（ without delving into detailed discussions of its functionality）, Face ID系统利用高性能摄像头来识别并捕获用户的面部特征（ recognize and capture facial features）。然而软件如何确定所捕捉到的是一个人的脸？换句话说，在这种情况下（ in such a scenario）, 如果软件误将一条腿当作脸部进行扫描（ scan legs instead of faces），将会导致严重的后果（ severe consequences）。
To ascertain if something constitutes a face, an algorithm may attempt to identify its mouth (or eyes or nose). Once such features are identified, it may classify it as depicting a face.This raises the question of how to distinguish between these features effectively.Additionally one might define that a mouth consists of two lines with specific orientation requirements and must not be positioned above the nose in order to accurately identify facial structures.These definitions can become increasingly complex over time and will need to be tailored for each sub-item
判断某物是否为一张脸时
Essentially, the central concept behind deep learning lies in the ability for the system to autonomously acquire intricate features without relying solely on raw data inputs. The algorithm learns this by being exposed to thousands of face images, subsequently developing a structured approach for facial recognition. The first step involves the identification of fundamental visual elements such as lines, edges, and corners. Building upon these basics, the system progresses to recognize more complex features like mouths and eyes. Finally, by integrating these components together, it can discern even higher-order characteristics such as facial hair or minor anomalies like dimples.
这一核心概念表明，在深度学习中仅通过分析原始数据便可提取出其中的功能。 该算法旨在利用大量面部图像样本训练模型，并在此基础上构建一个系统来逐步识别不同层面的 facial 特征。 系统可能首先致力于识别构成 facial 基本元素的部分——线条、边缘与角度等 low-level 特征。 在随后的学习阶段，则会关注于识别像 mouth、eyes 和 nose 这样的关键部位。 最后会整合这些基础信息以识别更为复杂的 feature 如 facial hair 或者 mole marks.
Image by Trist’n Joseph 图片由Trist'n Joseph
Deep learning is essentially a sophisticated process that allows computers to emulate human behavior by automatically extracting the most vital information required for making informed decisions. Data science has transformed numerous fields, largely attributed to cutting-edge technologies like AI and computing power. Real-world challenges are inherently intricate, which is why we should consider employing deep learning as a potent tool for addressing them.
可以说深度学习究竟是什么？它是一个强大的过程，在这个过程中计算机能够自动提取最有价值的信息，并为未来的决策提供依据，从而模拟人类的行为模式。数据科学正在彻底改变许多领域的原因之一就是依赖于AI和计算能力这些令人难以置信的强大工具。现实中许多问题都极为复杂，在这种情况下让我们尝试运用深度学习这一技术来寻找解决方案！
Deep learning represents a specialized area within the broader field of machine learning. It is characterized by the use of neural networks, which are composed of interconnected layers of artificial neurons. These neural networks are designed to model and simulate the way biological brains process information, enabling them to learn complex patterns and relationships in data. The concept of deep learning has revolutionized many areas of artificial intelligence, leading to advancements in natural language processing, computer vision, and autonomous systems. By leveraging large datasets and powerful computational resources, deep learning models have demonstrated remarkable success in solving previously intractable problems across diverse industries.
machinelearningmastery.com/what-is-deep-learning/
该资源的位置可访问于NN&DL
neuronetworksanddeeplearning.com/
mathworks.com/discovery/deep-learning.html
MathWorks Deep Learning Discovery Page
MathWorks Deep Learning Discovery Page
Deep Learning and Machine Learning Discoveries at MathWorks
Other Useful Material:
其他有用的材料：
This article explores the concept of deep learning and its underlying mechanisms. It provides a comprehensive overview of how neural networks operate through multiple layers to model and analyze complex data patterns. The text emphasizes the significance of deep learning in advancing artificial intelligence applications, highlighting its potential to transform various industries by enabling machines to learn and improve from experience. Additionally, it delves into the technical aspects, discussing the architecture of neural networks, activation functions, and optimization techniques that are fundamental to training effective deep learning models. By exploring these topics in depth, the article aims to provide readers with a thorough understanding of deep learning principles and their practical implementations.
这个方法在解决复杂问题方面表现非常出色。
www.youtube.com/watch?v=6M5VXKLf4D4
www.youtube.com/watch?v=6M5VXKLf4D4
https://www.inertia7.com/tristn
https://www.inertia7.com/tristn
翻译自: https://towardsdatascience.com/what-is-deep-learning-adf5d4de9afc
深度学习中的深度是指什么