吴恩达《深度学习》-课后测验-第三门课 结构化机器学习项目（Structuring Machine Learning Projects）-Week2 Autonomous driving (case study) (case study)( 自动驾驶 （案例研究）)

标签：但我   result   gre   lam   cts   eterm   com   cer   帮助   

Week2 Autonomous driving (case study) (case study)( 自动驾驶 （案例研究）)

\1. To help you practice strategies for machine learning, in this week we’ll present another scenario and ask how you would act. We think this “simulator” of working in a machine learning project will give a task of what leading a machine learning project could be like! (为 了帮助你练习机器学习策略，本周我们将介绍另一种场景并询问你将如何做。我们认为这个 在机器学习项目中工作的“模拟器”将给出一个引导机器学习项目的任务。)

You are employed by a startup building self-driving cars. You are in charge of detecting road signs (stop sign, pedestrian crossing sign, construction ahead sign) and traffic signals (red and green lights) in images. The goal is to recognize which of these objects appear in each image. As an example, the above image contains a pedestrian crossing sign and red traffic lights.( 你受雇于 一家创业的自动驾驶的创业公司。您负责检测图片中的路标（停车标志，行人过路标志，前方 施工标志）和交通信号标志（红灯和绿灯），目标是识别哪些对象出现在每个图片中。例如， 上面的图片包含一个行人过路标志和红色交通信号灯标志。)

Your 100,000 labeled images are taken using the front-facing camera of your car. This is also the distribution of data you care most about doing well on. You think you might be able to get a much larger dataset off the internet, that could be helpful for training even if the distribution of internet data is not the same.

You are just getting started on this project. What is the first thing you do? Assume each of the steps below would take about an equal amount of time (a few days).

(您的 100,000 张带标签的图片是使用您汽车的前置摄像头拍摄的，这也是你最关心的数据分 布，您认为您可以从互联网上获得更大的数据集，即使互联网数据的分布不相同，这也可能对 训练有所帮助。你刚刚开始着手这个项目，你做的第一件事是什么？假设下面的每个步骤将花 费大约相等的时间（大约几天）。)

【 】Spend a few days training a basic model and see what mistakes it makes. (花几天时间训 练一个基本模型，看看它会犯什么错误。)

【 】 Spend a few days checking what is human-level performance for these tasks so that you can get an accurate estimate of Bayes error. (花几天的时间检查这些任务的人类表现，以便能 够得到贝叶斯误差的准确估计。)

【 】 Spend a few days getting the internet data, so that you understand better what data is available. (花几天时间去获取互联网的数据，这样你就能更好地了解哪些数据是可用的。)

【 】 Spend a few days collecting more data using the front-facing camera of your car, to better understand how much data per unit time you can collect. (花几天的时间使用汽车前置摄 像头采集更多数据，以更好地了解每单位时间可收集多少数据。)

?

\2. Your goal is to detect road signs (stop sign, pedestrian crossing sign, construction ahead sign) and traffic signals (red and green lights) in images. The goal is to recognize which of these objects appear in each image. You plan to use a deep neural network with ReLU units in the hidden layers. For the output layer, a softmax activation would be a good choice for the output layer because this is a multi-task learning problem. True/False?( 您的目标是检测道路标志（停车标志、行人过路标志、前方施工标志）和交通信号（红灯和绿灯）的图片，目标 是识别这些图片中的哪一个标志出现在每个图片中。 您计划在隐藏层中使用带有 ReLU 单位 的深层神经网络。 对于输出层，使用 Softmax 激活将是输出层的一个比较好的选择，因为 这是一个多任务学习问题，对吗？ )

【 】 True (正确) 【 】 False(错误)

?

\3. You are carrying out error analysis and counting up what errors the algorithm makes. Which of these datasets do you think you should manually go through and carefully examine, one image at a time?( 你正在做误差分析并计算错误率，在这些数据集中，你认为你应该手动仔 细地检查哪些图片（每张图片都做检查）？)

【 】10,000 randomly chosen images (随机选择 10,000 图片)

【 】500 images on which the algorithm made a mistake (500 张算法分类错误的图片。)

【 】10,000 images on which the algorithm made a mistake (10,000 张算法分类错误的图 片。)

【 】500 randomly chosen images(随机选择 500 图片)

?

\4. After working on the data for several weeks, your team ends up with the following data: 100,000 labeled images taken using the front-facing camera of your car. 900,000 labeled images of roads downloaded from the internet. Each image’s labels precisely indicate the presence of any specific road signs and traffic signals or combinations of them.

【 】 True 正确 【 】 False 错误

?

5.The distribution of data you care about contains images from your car’s front-facing camera; which comes from a different distribution than the images you were able to find and download off the internet. How should you split the dataset into train/dev/test sets? (你所关 心的数据的分布包含了你汽车的前置摄像头的图片，这与你在网上找到并下载的图片不同。 如何将数据集分割为训练/开发/测试集?)

【 】Choose the training set to be the 900,000 images from the internet along with 80,000 images from your car’s front-facing camera. The 20,000 remaining images will be split equally in dev and test sets. (选择从互联网上的 90 万张图片和汽车前置摄像头的 8 万张图片作为训练 集，剩余的 2 万张图片在开发集和测试集中平均分配。)

【 】Mix all the 100,000 images with the 900,000 images you found online. Shuffle everything. Split the 1,000,000 images dataset into 600,000 for the training set, 200,000 for the dev set and 200,000 for the test set. (将 10 万张前摄像头的图片与在网上找到的 90 万张图片随机混合，使 得所有数据都随机分布。 将有 100 万张图片的数据集分割为：有 60 万张图片的训练集、有 20 万张图片的开发集和有 20 万张图片的测试集。)

【 】Choose the training set to be the 900,000 images from the internet along with 20,000 images from your car’s front-facing camera. The 80,000 remaining images will be split equally in dev and test sets. (选择从互联网上的 90 万张图片和汽车前置摄像头的 2 万张图片作为训练 集，剩余的 8 万张图片在开发集和测试集中平均分配。)

【 】Mix all the 100,000 images with the 900,000 images you found online. Shuffle everything. Split the 1,000,000 images dataset into 980,000 for the training set, 10,000 for the dev set and 10,000 for the test set. (将 10 万张前摄像头的图片与在网上找到的 90 万张图片随机混合，使 得所有数据都随机分布。将有 100 万张图片的数据集分割为：有 98 万张图片的训练集、有 1 万张图片的开发集和有 1 万张图片的测试集。)

?

\6. Assume you’ve finally chosen the following split between of the data:( 假设您最终选择了以 下拆分数据集的方式: )

You also know that human-level error on the road sign and traffic signals classification task is around 0.5%. Which of the following are True? (Check all that apply).(您还知道道路标志和交通 信号分类的人为错误率大约为 0.5％。以下哪项是真的（检查所有选项）? )

【 】 Your algorithm overfits the dev set because the error of the dev and test sets are very close.(由于开发集和测试集的错误率非常接近，所以你的算法在开发集上过拟合了。)

【 】 You have a large data-mismatch problem because your model does a lot better on the training-dev set than on the dev set(你有一个很大的数据不匹配问题，因为你的模型在训练-开发集上比在开发集上做得好得多。)

【 】 You have a large avoidable-bias problem because your training error is quite a bit higher than the human-level error.(你有一个很大的可避免偏差问题，因为你的训练集上的错误率比人为错误率高很多。)

【 】You have a large variance problem because your training error is quite higher than the human-level error.(你有很大的方差的问题，因为你的训练集上的错误率比人为错误率要高得 多。)

【 】 You have a large variance problem because your model is not generalizing well to data from the same training distribution but that it has never seen before.(你有很大的方差的问题， 因为你的模型不能很好地适应来自同一训练集上的分布的数据，即使是它从来没有见过的数 据。)

?

\7. Based on table from the previous question, a friend thinks that the training data distribution is much easier than the dev/test distribution. What do you think? (根据上一个问题的表格， 一位朋友认为训练数据分布比开发/测试分布要容易得多。你怎么看？)

【 】Your friend is right. (I.e., Bayes error for the training data distribution is probably lower than for the dev/test distribution.) (你的朋友是对的。 （即训练数据分布的贝叶斯误差可能低 于开发/测试分布）。)

【 】Your friend is wrong. (I.e., Bayes error for the training data distribution is probably higher than for the dev/test distribution.) (你的朋友错了。（即训练数据分布的贝叶斯误差可 能比开发/测试分布更高）)

【 】There’s insufficient information to tell if your friend is right or wrong. (没有足够的信息 来判断你的朋友是对还是错。)

?

\8. You decide to focus on the dev set and check by hand what are the errors due to. Here is a table summarizing your discoveries:

In this table, 4.1%, 8.0%, etc.are a fraction of the total dev set (not just examples your algorithm mislabeled). I.e. about 8.0/14.3 = 56% of your errors are due to foggy pictures. (在这个表格中， 4.1％、8.0％这些比例是总开发集的一小部分（不仅仅是您的算法错误标记的样本），即大约 8.0 / 14.3 = 56％的错误是由于雾天的图片造成的。)

The results from this analysis implies that the team’s highest priority should be to bring more foggy pictures into the training set so as to address the 8.0% of errors in that category. True/False? (从这个分析的结果意味着团队最先做的应该是把更多雾天的图片纳入训练集，以便解决该类别中的 8%的错误，对吗？)

【 】 True because it is the largest category of errors. As discussed in lecture, we should prioritize the largest category of error to avoid wasting the team’s time. (是的，因为它是错误率 最大的类别。正如视频中所讨论的，我们应该对错误率进行按大小排序，以避免浪费团队的时 间。)

【 】 True because it is greater than the other error categories added together (8.0 > 4.1+2.2+1.0). (是的，因为它比其他的错误类别错误率加在一起都大(8.0 > 4.1+2.2+1.0)。)

【 】 False because this would depend on how easy it is to add this data and how much you think your team thinks it’ll help. (错误，因为这取决于添加这些数据的容易程度以及您要考虑 团队认为它会有多大帮助。)

【 】 False because data augmentation (synthesizing foggy images by clean/non-foggy images) is more efficient. (错误，因为数据增强(通过清晰的图像+雾的效果合成雾天的图像)更 有效。)

?

\9. You can buy a specially designed windshield wiper that help wipe off some of the raindrops on the front-facing camera. Based on the table from the previous question, which of the following statements do you agree with? (如果合成的图像看起来逼真, 就好像您在有雾的天 气中添加了有用的数据来识别道路标志和交通信号一样。)

【 】 2.2% would be a reasonable estimate of the maximum amount this windshield wiper could improve performance. (对于挡风玻璃雨刷可以改善模型的性能而言，2.2％是改善的最大 值。)

【 】2.2% would be a reasonable estimate of the minimum amount this windshield wiper could improve performance. (对于挡风玻璃雨刷可以改善模型的性能而言，2.2％是改善最小 值。)

【 】2.2% would be a reasonable estimate of how much this windshield wiper will improve performance. (对于挡风玻璃雨刷可以改善模型的性能而言，改善的性能就是 2.2％。)

【 】2.2% would be a reasonable estimate of how much this windshield wiper could worsen performance in the worst case.( 在最坏的情况下，2.2%将是一个合理的估计，因为挡风玻璃刮 水器会损坏模型的性能。)

?

10.You decide to use data augmentation to address foggy images. You find 1,000 pictures of fog off the internet, and “add” them to clean images to synthesize foggy days, like this: (您决定使 用数据增强来解决雾天的图像，您可以在互联网上找到 1,000 张雾的照片，然后拿清晰的图 片和雾来合成雾天图片，如下所示：)

Which of the following statements do you agree with? (你同意下列哪种说法？（检查所有选 项）)

【 】So long as the synthesized fog looks realistic to the human eye, you can be confident that the synthesized data is accurately capturing the distribution of real foggy images (or a subset of it), since human vision is very accurate for the problem you’re solving. (只要合成的雾对人眼来 说看起来逼真，你就可以确信合成的数据和真实的雾天图像差不多，因为人类的视觉对于你正在解决的问题是非常准确的。)

【 】Adding synthesized images that look like real foggy pictures taken from the front-facing camera of your car to training dataset won’t help the model improve because it will introduce avoidable-bias. (将合成的看起来像真正的雾天图片添加到从你的汽车前摄像头拍摄到的图片的 数据集对与改进模型不会有任何帮助，因为它会引入可避免的偏差。)

【 】There is little risk of overfitting to the 1,000 pictures of fog so long as you are combing it with a much larger (>>1,000) of clean/non-foggy images. (只要你把它与一个更大（远大于 1000）的清晰/不模糊的图像结合在一起，那么对雾的 1000 幅图片就没有太大的过拟合的风 险。)

?

11.After working further on the problem, you’ve decided to correct the incorrectly labeled data on the dev set. Which of these statements do you agree with? (Check all that apply). (在进一步 处理问题之后，您已决定更正开发集上错误标记的数据。 您同意以下哪些陈述？ （检查所 有选项）)

【 】You should also correct the incorrectly labeled data in the test set, so that the dev and test sets continue to come from the same distribution (您还应该更正测试集中错误标记的数据，以 便开发和测试集来自同一分布。)

【 】You should correct incorrectly labeled data in the training set as well so as to avoid your training set now being even more different from your dev set. (您应该更正训练集中的错误标记 数据, 以免您现在的训练集与开发集更不同。)

【 】You should not correct the incorrectly labeled data in the test set, so that the dev and test sets continue to come from the same distribution (您不应该更正测试集中错误标记的数据，以便开发和测试集来自同一分布。)

【 】You should not correct incorrectly labeled data in the training set as well so as to avoid your training set now being even more different from your dev set. (您不应更正训练集中的错误标记的数据, 以免现在的训练集与开发集更不同。)

?

12.So far your algorithm only recognizes red and green traffic lights. One of your colleagues in the startup is starting to work on recognizing a yellow traffic light. (Some countries call it an orange light rather than a yellow light; we’ll use the US convention of calling it yellow.) Images containing yellow lights are quite rare, and she doesn’t have enough data to build a good model. She hopes you can help her out using transfer learning. (到目前为止，您的算法仅能识 别红色和绿色交通灯，该公司的一位同事开始着手识别黄色交通灯（一些国家称之为橙色光 而不是黄色光，我们将使用美国的黄色标准），含有黄色灯的图像非常罕见，而且她没有足 够的数据来建立一个好的模型，她希望你能用迁移学习帮助她。)

What do you tell your colleague? (你告诉你的同事怎么做？)

【 】She should try using weights pre-trained on your dataset, and fine-tuning further with the yellow-light dataset. (她应该尝试使用在你的数据集上预先训练过的权重，并用黄灯数据集 进行进一步的微调。)

【 】If she has (say) 10,000 images of yellow lights, randomly sample 10,000 images from your dataset and put your and her data together. This prevents your dataset from “swamping” the yellow lights dataset. (如果她有 10,000 个黄灯图像，从您的数据集中随机抽 取 10,000 张图像，并将您和她的数据放在一起，这可以防止您的数据集“淹没”她的黄灯数据 集。)

【 】You cannot help her because the distribution of data you have is different from hers, and is also lacking the yellow label. (你没办法帮助她，因为你的数据分布与她的不同，而且 缺乏黄灯标签的数据。)

【 】Recommend that she try multi-task learning instead of transfer learning using all the data. (建议她尝试多任务学习，而不是使用所有数据进行迁移学习。)

?

13.Another colleague wants to use microphones placed outside the car to better hear if there’re other vehicles around you. For example, if there is a police vehicle behind you, you would be able to hear their siren. However, they don’t have much to train this audio system. How can you help? (你已经在一个庞大的数据集上训练了你的模型，并且她有一个小数据集。 尽管您 的标签不同，但您的模型参数已经过训练，可以识别道路和交通图像的许多特征，这些特征 对于她的问题很有用。 这对于转移学习来说是一个完美的例子，她可以从一个与您的架构 相同的模型开始，改变最后一个隐藏层之后的内容，并使用您的训练参数对其进行初始化。 另一位同事想要使用放置在车外的麦克风来更好地听清你周围是否有其他车辆。 例如，如 果你身后有警车，你就可以听到警笛声。 但是，他们没有太多的训练这个音频系统，你能 帮忙吗？)

【 】Transfer learning from your vision dataset could help your colleague get going faster. Multi-task learning seems significantly less promising. (从视觉数据集迁移学习可以帮助您的 同事加快步伐，多任务学习似乎不太有希望。)

【 】Multi-task learning from your vision dataset could help your colleague get going faster. Transfer learning seems significantly less promising. (从您的视觉数据集中进行多任务学习可 以帮助您的同事加快步伐，迁移学习似乎不太有希望。)

【 】Either transfer learning or multi-task learning could help our colleague get going faster. (迁移学习或多任务学习可以帮助我们的同事加快步伐。)

【 】Neither transfer learning nor multi-task learning seems promising. (迁移学习和多任务 学习都不是很有希望。)

?

14.To recognize red and green lights, you have been using this approach: (要识别红灯和绿灯， 你使用这种方法：)

(A) Input an image (x) to a neural network and have it directly learn a mapping to make a prediction as to whether there’s a red light and/or green light (y). (将图像 xx 输入到神经网络， 并直接学习映射以预测是红灯(和/或)绿灯(y)。)

A teammate proposes a different, two-step approach: (一个队友提出了另一种两步走的方法：)

(B) In this two-step approach, you would first (i) detect the traffic light in the image (if any), then (ii) determine the color of the illuminated lamp in the traffic light. (在这个两步法中，您首 先要检测图像中的交通灯（如果有），然后确定交通信号灯中照明灯的颜色。)

Between these two, Approach B is more of an end-to-end approach because it has distinct steps for the input end and the output end. True/False? (在这两者之间，方法 B 更多的是端到端的方 法，因为它在输入端和输出端有不同的步骤，这种说法正确吗？)

?

15.Approach A (in the question above) tends to be more promising than approach B if you have a __ (fill in the blank). (如果你有一个__ ,在上面的问题中方法 A 往往比 B 方法更有效。)

【 】 Large training set (大训练集)

【 】Multi-task learning problem. (多任务学习的问题。)

【 】Large bias problem. (偏差比较大的问题)

【 】Problem with a high Bayes error. (高贝叶斯误差的问题。)

吴恩达《深度学习》-课后测验-第三门课 结构化机器学习项目（Structuring Machine Learning Projects）-Week2 Autonomous driving (case study) (case study)( 自动驾驶 （案例研究）)

标签：但我   result   gre   lam   cts   eterm   com   cer   帮助   

原文地址：https://www.cnblogs.com/phoenixash/p/12155464.html