Google AI certification
Axon (output)
C
Synapse (connection)
D
Autonomous car
Autonomous cars apply a wide range of techniques to function. These include statistics, robotics, and machine learning.
Which of the activations described above gives: the highest output for an input of 5?
Identity
Exercise 5 - A smaller rowboat
One of them is NNNN -> FNFN -> NNFN -> FFFN -> NFNN -> FFNF -> NFNF -> FFFF, and the other NNNN -> FNFN -> NNFN -> FNFF -> NNNF -> FFNF -> NFNF -> FFFF. (ANSWER IS 7)
What state should be in box 2?
State B [since box 1 contains state E, there are two possibilities for box 2: states B and F. Choosing state F in box 2 would lead to a dead-end at box 5, so the correct option must be state B. Also note that box 2 has two transitions to other states, which implies that it must be a state where the two discs are on top of each other.]
Who is the user most similar to Travis?
When you calculate the similarities between Travis and all the other users, Ville and Travis will have the largest similarity with a similarity of 3.
The weather forecast says it's going to rain with 0% probability tomorrow but the day turns out to be rainy.
Wrong [The weather forecast was wrong because a 0% probability means that it should definitely not rain. But it did.]
Now use the naive Bayes method to calculate the posterior odds for spam given the message "million dollars adclick conferences". You should again start with the prior odds 1:1, and then multiply the odds repeatedly by the likelihood ratios for each of the four words. Notice that the likelihood ratios are tabulated above for your reference (these are the numbers 5.1, 0.8, and so on). Your task: Express the result as posterior odds without any rounding of the result. You may take a look at the solution of the previous exercise for help.
The answer is 65.1168. We start in the same way as the previous exercise. Multiplying the prior odds by the likelihood ratio 5.1 (for the word 'million') gives posterior odds 5.1. Next we'll simply keep multiplying the odds by the likelihood ratios for the rest of the message. The likelihood ratios can be found in the table above: 0.8 ('dollars'), 53.2 ('adclick'), and 0.3 ('conference'). The product of all these numbers equals 1:1 × 5.1 × 0.8 × 53.2 × 0.3 = 65.1168. This means that for messages that include all these four words, there are on the average about 65 spam messages for each ham message, or about 651 spam messages for every 10 ham messages. If we wanted to get the probability value (which was not asked), it is about 651 / (651+10) = 651 / 661 or approximately 98.5 %. This message would probably end up in your junk mail folder.
Let's start with a message that only has one word in it: "million". Your task: Calculate the posterior odds for spam given this word using the table above, starting from prior odds 1:1. Keep in mind that the odds is not the same as the probability, which we would usually express as a percentage. Give your answer in the form of a single decimal number x.x using the dot '.' as the decimal separator. (Remember that odds can be represented as xx:yy or simply as a single decimal number, say z.z (where z.z = xx/yy). You may wish to revisit the discussion on this just before Exercise 9 in Section 3.1 (Odds and Probability).)
The correct answer is 5.1. As you may have noticed, the structure of this exercise is identical to that of the previous exercise about medical diagnosis. We have the class label spam or ham, and one piece of evidence that we can use to update our prior odds to obtain the posterior odds. We decided above that the prior odds are 1:1. The likelihood ratio is obtained by dividing the probability of the word 'million' in spam divided by the probability of the word 'million' in ham. This we already calculated above, and it can be found in the table of likelihood ratios: the value is 5.1. Now multiply the prior odds by the likelihood ratio to get 1:1 × 5.1 = 5.1. This is the posterior odds. Again, the posterior odds means that for messages that include the word 'million', there are on the average 5.1 spam messages for every ham message. Or to use whole numbers, there are 51 spam messages for every 10 ham messages. The probability value is therefore 51 / (51+10) = 51/61, or approximately 83.6 %.
It is probably less than 90
The few data points that we have make it impossible say almost anything about the life expectancy only based on the data. Of course, one can know a great deal about life expectancy from other sources but the data in the above chart is insufficient to do so. The first choice is clearly stating too much. While the intervals in the second and the third choice are likely to be valid, the word 'certainly' makes them unjustified. There is a chance, greater than zero, that the value turns out to be, for example, greater than 70. Thus the only choice that we can be comfortable with is the fourth one.
Probably between 50 and 90 years
The first choice would clearly be an odd estimate since the data strongly suggest that very few countries have life expectancy less than 50, and none of the data points with more than 12 years of education fall below 50. We can't be sure, of course, but life expectancy between 45 and 50 years would in this case be highly unexpected. The second choice is correct because it fits the general trend, and all data points with more than 12 years of education fall within this interval. The interval 69 to 71 years in the third choice could well include the actual value, but based on the above data, it would be too bold to claim to know the outcome with such high accuracy. The interval 15 to 150 years of the fourth choice would almost certainly include the actual value, but we think that it would be a poor summary of what we can learn from the data for the reason that it is too vague.
Consider the above breast cancer scenario. An average woman takes the mammograph test and gets a positive test result suggesting breast cancer. What do you think are the odds that she has breast cancer given the observation that the test is positive? First, use your intuition without applying the Bayes rule, and write down on a piece of paper (not in the answer box below) what you think the chances of having breast cancer are after a positive test result. The intuitive answer will not be a part of your answer. It will be just for your own information. Next, calculate the posterior odds for her having breast cancer using the Bayes rule. This will be your answer. Hints: Start by calculating the prior odds. Determine the probability of the observation in case of the event (cancer). Determine the probability of the observation in case of no event (no cancer). Obtain the likelihood ratio as the ratio of the above two probabilities. Finally, multiply the prior odds by the likelihood ratio. Enter the posterior odds as your solution below. Give the answer in the form xx:yy where xx and yy are numbers, without simplifying the expression even if both sides have a common factor.
The prior odds describe the situation before getting the test result. Since five out of 100 women have breast cancer, there is on the average five women with breast cancer for every 95 women without breast cancer, and therefore, the prior odds are 5:95. The likelihood ratio is the probability of a positive result in case of cancer divided by the probability of a positive result in case of no cancer. With the above numbers, this is given by 80/100 divided by 10/100, which is 8. The Bayes rule now gives the posterior odds of breast cancer given the positive test result: posterior odds = 8 × 5:95 = 40:95, which is the correct answer. So despite the positive test result, the odds are actually against the person having breast cancer: among the women who are tested positive, there are on the average 40 women with breast cancer for every 95 women without breast cancer. Note: If we would like to express the chances of breast cancer given the positive test result as a probability (even though this is not what the exercise asked for), we would consider the 40 cases with cancer and the 95 cases without cancer together, and calculate what portion of the total 40 + 95 = 135 individuals have cancer. This gives the result 40 out of 135, or about 30%. This is much higher than the prevalence of breast cancer, 5 in 100, or 5%, but still the chances are that the person has no cancer. If you compare the solution to your intuitive answer, they tend to be quite different for most people. This demonstrates how poorly suited out intuition is for handling uncertain and conflicting information.
Look at the game tree starting from the below board position. Using a pencil and paper, fill in the values of the bottom-level nodes where the game is over. Note that this time some of the games end in a draw, which means that the values of the node is 0 (instead of -1 or 1). Next continue filling the values of the nodes in the next level up. Since there is no branching at that level, the values on the second-lowest level are the same as at the bottom level. On the second-highest level, fill in the values by choosing for each node the maximum of the values of the child nodes - as you notice, this is a MAX level. Finally, fill in the root node's value by choosing the minimum of the root node's child nodes' values. This is the value of the game. Enter the value of the game as your answer.
The value is -1. The values on the second level are 0, 0, and -1. The values on the third level are -1, 0, -1, 0, -1, -1, which are the same as the values on the bottom level. As you can see, Max has all the reason to be serious since by playing in the bottom-right corner, Min can guarantee a win. The inevitable victory of Min can also be seen from the value of the game -1.
Suppose you monitor a weather forecaster for a long time. You only consider the days for which the forecast gives 80% chance of rain. You find that in the long run, on the average it rains on three out of every five days.
Wrong [The weather forecasts are wrong if they predict 80% chance of rain and it rains only 60% (three out of five) of the time in the long run. (Note that we'd really need to keep track of the accuracy for a long time to reach this conclusion but that's what "in the long run" means.) In practice, weather forecasters actually tend to provide this kind of 'wrong' predictions just to be safe: people are often quite disappointed when the weather turns out to be worse than predicted but pleasantly surprised when it turns out better than predicted.]
What kind of articles (in newspapers and magazines or other popular science outlets such as blogs, ...) are being written about AI - and do you think they are realistic? Do an online search about AI related to one of your interests. Choose one of the articles and analyze it. Mention the title of the article along with its author and where it was published (as a URL if applicable) in your answer. Explain the central idea in the article in your own words using about a paragraph of text (multiple sentences.) Based on your understanding, how accurate are the AI-related statements in the article? Explain your answer. Are the implications (if any) realistic? Explain why or why not.
1. Article Title ~ Q&A: Vivienne Sze on crossing the hardware-software divide for efficient artificial intelligence Author ~ Kim Martineau | MIT Quest for Intelligence URL ~ https://news.mit.edu/2021/vivienne-sze-crossing-hardwire-software-divide-efficient-ai-0428 2. The article is about AI obviously, but it has common questions about AI and answers for them that she answered. It also talks about one of Vivienne Sze's achievements (video compression for smartphones without draining the battery - we use these same video compression standards today). 3. The AI-related statements are very accurate because it came from a professional, and also it talks about what they are doing currently. I don't think that putting AI on a smartphone is very marketable but apparently Sze thinks so. Redesigning AI hardware is realistic because that is something that is possible.
How do you see AI affecting you in the future, both at work and in everyday life? Include both the positive and possible negative implications.
A companion that you can use in your home and maybe put into a biochip that you connect to your brain. AI will open the world for human advancements in technology. It will make most tasks easier and done with little to no effort at all. But of course, some people will use these technologies fully and forget about personal health. They will become chubby overall and the human race will be obese. But for the most part, it will improve many lives and help multiple avenues on many scales.
Do you think that filter bubbles are harmful? After all, they are created by recommending content that the user likes. What negative consequences, if any, may be associated with filter bubbles? Feel free to look for more information from other sources. Think of ways to avoid filter bubbles while still being able to recommend content to suit personal preferences. Come up with at least one suggestion. You can look for ideas from other sources, but we'd like to hear your own ideas too!
Filter bubbles are more helpful than they are harmful. They both benefit the user and the companies that are suggested to the user. There are also negative consequences with or associated with filter bubbles; such as, you may miss out on some important news that would impact you or others, Some people claim that it creates an identity for you--which can be a disadvantage as people may think that Filter Bubbles are almost categorizing us, and you won't see a broader scale of searches--instead only seeing your most searched up things instead of things you may like--but haven't searched up that much. As far as news and stuff like that they should be unfiltered to everyone or at least be so that the user can customize what type of news they want to see. Items can be suggested to the user with filter bubbles as much as desired because to me that really doesn't matter.
A spreadsheet that calculates sums and other pre-defined functions on given data
No (not AI) [The outcome is determined by the user-specified formula, no AI needed.]
The odds for rain in Helsinki are 206:159.
Previously we had the probability as 206/(206 + 159) = 206/365, which gives us roughly 0.5644, which rounds to 56.4%.
The odds for rain in San Diego are 23:342.
Previously we had the probability as 23/(23 + 342) = 23/365, which gives us roughly 0.0630, which rounds to 6.3%.
Steering a rocket into orbit
Robotics. (In order to steer a rocket into orbit robotics are needed to fire the engines at the right times and with the right power.)
Where would you put deep learning? (Deep learning is a part of machine learning.)
Section D
What state should be in box 6?
State A [Since box 4 contains state D, there are two possibilities for box 6: states A and C. Choosing state C would lead to a dead end in box 5, so the correct choice must be state A. Also note that box 6 has two transitions to other states, which implies that it must be a state where the two discs are on top of each other.]
What state should be in box 4?
State D [State D is the only option that is reachable from the right box on the second row.]
What state should be in box 1?
State E [State E is the only option that is reachable from the left box on the second row.]
What state should be in box 3?
State F [Since box 1 contains state E, there are two possibilities for box 3: states B and F. Choosing state B would lead to a dead end in box 5, so the correct choice must be state F.]
Online ad optimization
Statistics and Machine Learning. (In order to optimize ads online, machine learning and statistics are needed to deliver the correct type of ads to the right audience, and to measure the effectiveness of the optimization.)
Apply the Bayes rule to calculate the posterior odds for rain having observed clouds in the morning in Helsinki. As we calculated above, the prior odds for rain is 206:159 and the likelihood ratio for observing clouds is 9. Give your result in the form of odds, xx:yy, where xx and yy are numbers. (Note that xx and yy does not mean that the numbers should have two digits each.) Remember that when multiplying odds, you should only multiply the numerator (the xx part). For example, if you multiple the odds 5:3 by 5, the result is 25:3. Give the answer without simplifying the expression even if both sides have a common factor.
The answer is 1854:159. The prior odds are 206:159. The likelihood ratio is 9, so we get the posterior odds for rain given clouds to be 9 × 206:159 = 1854:159. So in the long run, on the days when we observe clouds in the morning, we can expect 1854 rainy days for every 159 rainless days, which is about the same as 12 rainy days for one rainless day. If we wanted to express this as a probability (even though this was not the question), we could use the formula x / (x+y) to get the value 1854 / (1854+159) which is about 0.92, or about 92% chance of rain when there are clouds in the morning. Better take an umbrella.
10-11 hours
8-9 hours of studying gives you roughly a 70% chance of passing. To have an 80% chance of passing, you should study for around 10-11 hours.
Dendrite (input)
B
the lowest output for an input of -5?
Identity
The odds for getting three of a kind in poker are about 1:46.
Previously we had the probability as 1/(1+ 46) = 1/47, which gives us roughly 0.0213, which rounds to 2.1%.
What state should be in box 5?
State C
Cell body
A
The weather forecast says it's going to rain with 90% probability tomorrow but the day turns out to be all sun and no rain.
Cannot be concluded [We can't conclude that the weather forecast was wrong based on only the single event. The forecast said it's going to rain with 90% probability, which means it would not rain with 10% probability or in one out of 10 days. It is perfectly plausible that the day in question was the 1 in 10 event. Concluding that the probability 90% was correct would also be wrong because by the same argument, we could then conclude that 80% chance of rain was also correct, and both cannot be correct at the same time.]
A GPS navigation system for finding the fastest route
Kind of/no/yes [the signal processing and geometry used to determine the coordinates isn't AI, but providing good suggestions for navigation (shortest/fastest routes) is AI, especially if variables such as traffic conditions are taken into account.]
Big data storage solutions that can store huge amounts of data (such as images or video) and stream them to many users at the same time
No [Storing and retrieving specific items from a data collection is neither adaptive or autonomous.]
Where would you put data science? (Data science needs computer science and AI. However, it also involves a lot of statistics, business, law, and other application domains, so it is usually not considered to be a part of computer science.)
Section E
the highest output for an input of -2.5?
Sigmoid
What is the predicted purchase for Travis?
Since Ville's latest purchase was sunscreen, we will recommend it also to Travis.
Style transfer filters in applications such as Prisma that take a photo and transform it into different art styles (impressionist, cubist, ...)
Yes [Such methods typically learn image statistics (read: what small patches of the image in a certain style look like up close) and transform the input photo so that its statistics match the style, so the system is adaptive.]
Photo editing features such as brightness and contrast in applications such as Photoshop
kind of & no [Adjustments such as color balance, contrast, and so on, are neither adaptive nor autonomous, but the developers of the application may use some AI to automatically tune the filters.]
Predicting the stock market by fitting a curve to past data about stock prices
Kind of/no/yes [Fitting a simple curve is not really AI, but there are so many different curves to choose from, even if there's a lot of data to constrain them, that one needs machine learning/AI to get useful results.]
Philosophy of AI
The definition of AI I like best is that AI is an autonomous and adaptive system. The reason why I think this is the best definition is that it covers most of the AI capabilities that we are discovering today. I would define AI as an autonomous adaptive system that is used to do things that things on the internet do with very little human intervention.
The odds for rain in Helsinki are 206:159.
There are 206 rainy days for 159 dry days, so the probability is 206/(206+159) = 206/365.
A music recommendation system such as Spotify that suggests music based on the users' listening behavior
Yes [The system learns from the users' (not only your) listening behavior.]
C
80 + 4 = 84
B
80 - 8 + 1 = 73
10.0 + 5.4 × 8 + (-10.2) × 5 + (-0.1) × 22 + 101.4 × (-5) + 0.0 × 2 + 12.0 × (-3) = -543.0 What are the inputs? a) 8, 5, 22, -5, 2, -3 b) 5.4, 8, -10.2, 5, -0.1, 22, 101.4, -5, 0.0, 2, 12.0, -3 c) 5.4, -10.2, -0.1, 101.4, 0.0, 12.0 d) 43.2, -51.0, -2.2, -507.0, 0.0, -36.0
A
10.0 + 5.4 × 8 + (-10.2) × 5 + (-0.1) × 22 + 101.4 × (-5) + 0.0 × 2 + 12.0 × (-3) = -543.0 What happens when the fifth input is incremented by one? a) nothing b) the output increases by one c) the output increases by two d) something else
A
A
A: 80 - 5 + 6 = 81
10.0 + 5.4 × 8 + (-10.2) × 5 + (-0.1) × 22 + 101.4 × (-5) + 0.0 × 2 + 12.0 × (-3) = -543.0 What is the intercept term in the expression? a) 543.0 b) 10.0 c) -3 d) 5.4?
B
In the United States presidential election 2016, a well-known political forecast blog, Five-Thirty-Eight, gave Clinton a 71.4% chance of winning (vs Trump's 28.6%). However, contrary to the prediction, Donald Trump was elected the 45th president of the United States.
Cannot be concluded [Cannot be concluded to be wrong (or right). Sometimes unlikely things happen. Considering the previous item, it would actually have been wrong to predict, say, 90% or 100% chance for Trump if there simply isn't enough information available to anticipate the outcome. In other words, perhaps Trump's victory had a rare (or rareish) event with 28.6% probability. Such events are expected to happen in more than one out of four cases, after all.]
10.0 + 5.4 × 8 + (-10.2) × 5 + (-0.1) × 22 + 101.4 × (-5) + 0.0 × 2 + 12.0 × (-3) = -543.0 Which of the inputs needs to be changed the least to increase the output by a certain amount? a) first b) second c) third d) fourth
D
Customer service chatbot
Machine learning (A customer service chatbot will need machine learning to process human produced language in such a way that it can act on it.)
For this exercise, we want you to think about how AI is portrayed. Do an online image search for the term "AI" and see what kinds of pictures come up. If you are using Google search, you should choose "Images" at the top of the screen. What's the general impression you get about AI from the image search results? Is this an accurate representation of AI? Why or why not?
Most of the images are either robots or images about a 3D constructed mind. There are also pictures that can be depicted as AI being our future advancements. There really is no one correct version of AI. AI takes multiple forms and ideas. So yes, I do think these are accurate representations of AI.
Where would you put computer science? (Computer science is a relatively broad field that includes AI but also other subfields such as distributed computing, human-computer interaction, and software engineering.)
Section A
Where would you put AI? (AI is a part of computer science.)
Section B
Where would you put machine learning? (Machine learning is usually considered to be a part of AI)
Section C
Summarizing gallup results
Statistics (Summarizing gallup results is a classical case of using statistics to produce insights.)
The odds for rain in San Diego are 23:342.
There are 23 rainy days for 342 dry days, so the probability is 23/(23+342) = 23/365.
The odds for getting three of a kind in poker are about 1:46.
There are 46 situations where you do not get three of a kind for one where you get it, so the probability is 1/(1+46) = 1/47.
