Some questions asked previously during the Q&A time

Below are some questions that were asked during previous oral presentations.

General Questions

  • What is the primary goal of your project?
  • Why did you choose this topic?
  • What motivated you to do this project [or a certain part of the project]?
  • Which part of your project took the most time? (each person be prepared to answer)
  • What was the most difficult part of your project? (each person be prepared to answer)
  • Which person in your group was responsible for each part of your system?
  • How heavy was the workload?
  • Can you please explain [some particular] diagram?
  • Why did you have a problem with [some problematic part]?
  • Is there any way to get around that problem?
  • Did you need to do a lot of programming?
  • Was your focus on developing new algorithms or modifying existing ones?
  • Why did you choose this method (or these methods)?
  • If you did it another way, what results would you expect?
  • Can you think of one part that can be changed? If you could start over and do the project again, what would you do differently?
  • What percentage of your time did you spend on the main tasks?
  • How did you cooperate on this project?

FYT Questions

  • How realistic are these models?
  • When you combined the two methods, how did you weight the contribution of the two, i.e., which one was more important?
  • What is the purpose of this approach?
  • How is this different from the traditional approach?

Software Development Questions

  • What is the purpose of your system?
  • What group of people would benefit from using your system?
  • How does your software help the user?
  • How many users can your system accommodate at one time?
  • Does this system really work? What is the success or accuracy rate?
  • How is your system different from [some similar real system]?
  • How many lines of code did you write altogether?
  • What software tools did you learn through the project?
  • How do you know if you implemented the method [or algorithm] correctly?
  • What kind of information can you display?
  • How could you make the system more user-friendly?
  • How could you speed up the algorithm?
  • Can you demonstrate your system with another set of data?
  • When did you add the administrative part?
  • If you had an unlimited budget, what interesting conditions would be possible, i.e., what features or techniques would you add?
  • Are there any cases where one method does better than another?
  • Can you explain the normal flow from the user's point of view?
  • Will the system work on a mobile device?
  • Is there a DBMS on the server?
  • Can the user save the results?

Software Testing Questions

  • How did you test the system?
  • How many times did you test the system?
  • How did you confirm that the output is accurate?
  • How could you have done better testing?
  • What can you do to reduce the error?
  • How easy is it to learn?
  • How do you get the data?
  • What libraries did you use, and what code did you write on your own?
  • How many volunteer testers did you have? Did they provide feedback?

Machine Learning & Knowledge Discovery Questions

  • Where did you get your dataset? How large is it?
  • Was cleaning the data time consuming? Can you explain why?
  • Why was your error rate so low when you used this algorithm?
  • Did you do any cross validation?
  • What portion of the dataset was for training and what part was for testing accuracy?
  • Why was one algorithm faster than another?
  • How many items were in your sample data stream?
  • Why did one algorithm use more memory than another?
  • Can you show a graph for the memory usage to data size ratio?
  • Why didn't you use any deterministic algorithms?
  • Did you measure the discounted cumulative gain (DCG)?
  • Can you explain the idea behind the formulas for location filtering and activity filtering?
  • What factors affect accuracy?
  • What benefit does machine learning provide for your users?
  • What benefit does collaborative filtering provide?
  • How do you generate recommendations?
  • How do you know what users want?
  • How did you transfer [software in one language] to [another language]?
  • What do the X and Y axes represent?
  • If new data keeps coming in, how can you incorporate it?
  • What differences would you expect if you used a different dataset?
  • Would the technical part change with a different type of dataset?
  • How do you use GPS?
  • Can you explain more about [a certain tool you used]?
  • What's unique about [a certain tool you used]?

Mobile App Questions

  • What language did you use?
  • What platform did you use?
  • Can you describe the communication between the client and the server?
  • Do you synchronize the client and the server database?
  • How many simultaneous users can the server handle?
  • How do you utilize the accelerometer and gyroscope?
  • How much [more] time would it take to write a native iOS / Android app?
  • What data do you collect from the users?
  • How many data points?
  • How do you ensure the quality of the content shared by users?
  • How do you deal with users who share bad data?
  • Can you display PDF files?
  • If there was no API, did you try HTML parsing?
  • Can you explain more about the difference between the two modes?
  • Would it be difficult to provide support for other languages?

Networking and Localization Questions

  • Did you try GPS instead of Wi-Fi?
  • Why did you use [a certain] method?
  • What were the difficulties you faced in the porting process?
  • Where did you get your RSSI distance equation?
  • Was the RSSI at one meter a single measurement or an average of several measurements?
  • What did you do when the signal was not stable?
  • What distance did you use in a big room?
  • How dense did the reference nodes need to be?
  • Have you tested the network settings with furniture in the room?
  • How did you train the system with sample data?
  • How could you improve the localization?
  • How did you know the routing information?
  • Did you use different devices to collect signal strength data? If so, what was the difference in performance?
  • Is there any problem if the robot calibration changes from day to day?
  • Why did you choose a light sensor and an accelerometer?
  • If you did this demo in dim light, would it affect the performance?
  • If the room had a window, would it make a difference?
  • If you block the light, will the system be affected?
  • Since it was very tedious work to collect signal strength data, would it have helped to program the robot to do some of the work?
  • What was the level of accuracy you used to determine the best grid cell size?
  • Why did it work better indoors than it did outdoors?
  • Is the localization affected by the door being open or closed?
  • If the grid cell size is larger, how does that affect the accuracy?
  • Do you always keep a recursive average?
  • Did you ever think about using multiple sensors to receive signals?
  • If you were to use multiple sensors to receive signals, how would you deal with collisions?
  • Are you limited to using a fixed route scenario?
  • Is the delay setting an ad hoc value (i.e., designed just for this problem), or is there another reason for it?
  • Have you thought about privacy issues?
  • How can you make sure none of the owner's private information is disclosed?
  • How do you address user privacy with the accelerometer and the light sensors?

Prediction-related Questions

  • How accurate is your prediction?
  • Did you make past performance an indicator for future performance?
  • When you calculated your error, did you use historical data?
  • Did you simply predict things that already happened or use real incoming data?
  • What accuracy do you get when you predict future prices?
  • Can you update the number of days for prediction in the future?
  • What is the maximum number of days for future price prediction?
  • Does the pattern flatten out after 100 days?
  • Would you use your own money to invest with this system?
  • Which prediction method yielded the best results?
  • How does a Naive Bayes (non-independent) classifier affect the output?
  • Why does the prediction get worse when using data mining?
  • How did you do the analysis?
  • What if the trends are contradictory?
  • Is [a certain method of prediction] easy or hard?
  • Did you read the UST PhD student's related work?
  • How do you rotate the terrain (in your simulation)?
  • How was running time affected by the size of the cells?
  • How many grid cells did you use, and what size were they?
  • What is the distance between objects or exhibits?
  • What patterns did you look for?
  • Who invented that algorithm?
  • Does your system produce better or worse results than other methods?
  • How far ahead can you make predictions?

Vision and Graphics Questions

  • Is there no correlation in this chart?
  • Is there correlation with the index?
  • How accurate is your system?
  • Can you show sample data?
  • Where did you get the data? Is it real data?
  • When you draw tiles, do you do [some procedure] in advance?
  • Why does it take up so much memory if you break it into tiles?
  • Does the graph include the image?
  • Did you consider parallel programming for the tile generator?
  • Would it work if you get the source from all of them at the same time?
  • Have you thought about using homography?
  • Why did you rely on mapping? 
  • How do you map the the two parts together in real-time? 

RFID Sensing Questions

  • What if the size of the scanned objects changes?
  • What did you think of [a certain] equation?
  • Can you explain why [a certain] equation was not accurate enough?
  • Is there any chance that your co-efficient is not accurate enough?
  • Upon what are your transmitted gain and received gain based on?
  • Did you count measurement errors?
  • Did the measurement errors significantly affect the calculations?
  • Are you sure the angles on the simulated system are the same as the angles on the actual antenna?
  • What happens when an RFID tag is detected by more than one antenna?
  • Do the virtual tags determine granularity?
  • In your experiment, why did you use 3 tags at the same point?
  • If there is interference, why do you use the tags concurrently?
  • Why didn't you place the tags sparsely?

Music and Pattern Matching Questions

  • Can you do it by either singing or humming?
  • Can you hear the pattern of popular songs?
  • Can you hear the pattern of unpopular songs?
  • When you check for a pattern match, which part of songs do you check and how do you do it?
  • If the range of one is 140, and the range of the other is 360, then how are they similar?
  • How long is a song (time, number of notes)?
  • Can you change it to 5 seconds instead of 10?
  • Is your data pre-computed and stored in a database?
  • How do you determine K in the equation?
  • How does the center change?
  • Is the sampling rate for the voice?
  • Why did you have such a high sampling rate?
  • Voices don't go over a sampling rate of 1100, do they?
  • What software did you use to convert the MIDI files?
  • Do you convert the MIDI files to text format?

Shortest Path Algorithm Questions

  • Did you make the images on the interface?
  • How long does it take to pre-compute the shortest paths?
  • When do you pre-compute the shortest paths?

Web Searching Questions

  • Did you train the system with some sample data?
  • How does the system recognize images?

Game Questions

  • Can I see your algorithm for the AI?
  • How does the AI consider players' health points, magic points and distance when targeting them?
  • Does a player who stays out of the way in a corner have a better chance of surviving?
  • Does a player gain points by killing an AI entity?
  • Did you use an existing game engine or write your game from scratch?
  • How many lines of code did you use to write the prototype?
  • Can you explain what the player gets if he finishes a level, defeats a boss, wins the game, etc.?
  • What is the value of collecting pets or monsters?
  • What is the point of having mini games?

Corporate Software Questions

  • Can you cite an example of the two companies' different requirements?
  • Do you have two versions of your system?
  • What does it mean to improve the system's circulation and richness?
  • If a user fails to enter the correct password three times, is it possible to break into the system?
  • Why is it necessary to reset the password?
  • Is the system live now?
  • Was it necessary to modify either database for the sake of compatibility?
  • How did you do the database translation?
  • Did you have to translate the database often?
  • Did you do the conversion manually?
  • It seems easy to delete some companies. Will that cause a problem?
  • How will deleting a company affect reports?
  • Technically, what happens when you delete a company?
  • Do you allow cascade deletions?

Embedded Software Questions

  • What sensor are you using?
  • How do you use RFID?
  • What if there is an obstacle?
  • What if the robot gets out of line?
  • How do you re-align it?
  • Why did you use an LED light and not a laser?

Probabilistic Database Questions

  • Do you have to find all the vertices with the functional dependency (FD) cleanup method?
  • Did you develop the FD normal form? Why?
  • Can you give examples of how the FD normal exists?
  • How is this different from traditional FD?
  • Are the FD and normalization algorithms related?
  • If I am interested in the efficiency issue, how do I enforce integrity constraints and inclusion constraints?
  • Do you apply your algorithms to the FD problem?
  • Why wouldn't you use an index?
  • What have you done to check for FD?
  • Does the distance depend on the probability?



Copyright HKUST CSE Dept. 2024
Blog template built for Bootstrap by @mdo.
Back to top