Organizational Questions:
Q: How many questions are in the exam?
A: About 15 questions.
Q: Could you tell us more about the exam? It is just for 1h?
A: Yes, 1h.
Q: Would the exam be based on the Übung or is it completely based on lecture PDFs?
A: It is mostly based on the tutorials. But it is beneficial to know the lecture PDF content, too.
Q: Could you repeat the 5 topics of the exam again please?
A: Information Theory, Compression Techniques, Image Encoding, Video Encoding, Audio Encoding.
Q: Ist die Klausur immer noch wie vorher, ähnlich wie die Klausur in Opal?
A: Yes, it's similar.
Q: Is it an open book exam?
A: No, it's closed book.
Q: Calculator is not allowed? How can we calculate Log or related queries without calculator?
A: All calculation tasks are solvable manually. There will not be hard to calculate float numbers to handle.
Q: If questions like Huffman Encoding is asked, will only 1 hour be enough to solve the whole exam?
A: Yes, however it's good to know the algorithmic steps by heart, so that you can quickly execute the algorithm, without losing too much time thinking about how it works.
Q: Can we use pen and paper during exam?
A: Yes, a blank sheet of paper is allowed.
Q: Do we need to use phone camera/webcam during exam? Do we need 2 cameras?
A: Yes. Camera 1 from your notebook and camera 2 from your smartphone which shows you sitting in front of your computer from the side.
Q: How will we get the link to join the exam?
A: You will receive an Opal email from us later today.
Q: Are multiple choice questions included in the exam?
A: Yes, a few, but not many.
Q: How big is the practical part? Or is it more theoretical?
A: It's a mix of both.
Q: Are the gray areas in the lecture important?
A: Yes.
Q: Do we need 2 cameras? I have only camera from a smartphone, since I dont use a notebook.
A: Normally, yes. But in this case, your smartphone camera is mandatory.
Q: Any mock test available?
A: Yes, there is the exam from summer semester 2019 available in the Opal course. https://bildungsportal.sachsen.de/opal/auth/RepositoryEntry/3190226946/CourseNode/100097323611697
Q: What are the total exam marks?
A: About 30 to 35 points.
Q: How much needed to pass exam?
A: 50% of the points.
Q: Is there negative marking?
A: In general no, only in multiple choice tasks, but there the grading system will be precisely specified for you to see.
Q. Do we have to upload answers, e.g. a picture like for Huffman or Shannon-Fano during the exam?
A: No, there are no upload tasks.
Q: Will the Huffman or Shannon-Fano task be taken from the tutorial or will it contain other, changed values?
A: Yes, there will be other values contained.
Q: In case of complex calculations, what to do? Can we explain it in sentence or words form?
A: There won't be any complex calculations. Calculations tasks are designed to contain friendly numbers only, which are easy to process.
Q: For the german speakers: If we don't know the german word for something, can we also use the english one?
A: Yes.
Q: Can the whole class see our cameras?
A: No.
Q: How much time would the exam correction take?
A: You can expect the results by the end of August.
Q: When do we have to register at maxexam?
A: Sometime between today, as soon as you get the link, and Monday at the latest.
Q: Is the exam in each room same?
A:Yes.
Q: Can we answer the questions in our own words or do we have to write the exact wording as per the lectures?
A: Yes, you can write in your own words.
Subject-related Questions:
Q: What would be the solution for Hamming distance: 11111 ^ 00000 ^ 10101?
A: It can only be calculated among 2 code words. Here, you would need to calculate 3 combinations.
Q: Is it 3, 2, 3 ? for the above question?
A: No, it's 5, 2, 3.
Q: Could you explain why only 3, since it compares seperately 3 times?
A: Yes, you calculate code word 1 vs. 2, code word 1 vs. 3 and code word 2 vs. 3.
Q: Do we have to know the equations about quantizers, (P)SNR and color conversion (RGB to CMYK or to HSV)?
A: Quantizers and (P)SNR, yes. Color conversion equations, no. Entropy, Information content yes. error metrics in tutorial 1 or 2.
Q: What are Additive and subtractive color spaces?
A: Please see lecture PDF.
Q: In motion vectors, are logarithmic search calculation related problems asked or not?
A: No calculations are asked. We might ask you to explain the principle or how different search strategies compare to each other.
Q: Could you please explain mid-tread and mid-rise quantizers?
A: The general information is explained in the lecture and tutorial. If questions are left, be more precise, what exactly you do not understand.
Q: I have heard, that if we lose one part of the question, we will lose all of the points, is it right?
A: No. Each piece of information that we ask for is graded individually.
Q: Will there be drag and drop questions?
A: No.
Q: Can you please repeat the Huffman Encoding and draw a tree, please?
A: Please have a look at slide 14 of the Compression Techniques tutorial PDF, there you will find an animated step-by-step solution of the tree construction.
Q: Huffman encoding: It can happen, that I put some of the elements on the right or left, would that make a difference? Because on Opal with auto grading, it would be considered false when I change the order if they are left or right.
A: We will have a look manually and make sure that we grade your tree
Q: Do we need to draw diagrams in the exam?
A: No.
Q: Frame sequence in MPEG encoding.
A: Please see slide 12 from the Video Encoding lecture PDF: https://bildungsportal.sachsen.de/opal/FolderResource/3190226946/Unit_08_VideoEncoding_Slides.pdf
Q: Tutorial 6, task 2 solution is missing, quantization tables will be used for JPEG -Compression. Can you please give the solution or explain it here?
A: It's not missing. Please see here: https://iris1.informatik.tu-chemnitz.de/index.php/s/Er3RZZLgx884r5X, slide 5.
Q: The cycle of a traffic light is configured as follows: green (50 sec) → red (37,5 sec) → yellow (12,5 sec) → green. Construct a suitable, redundand 3 Bit encoding of the lights, explain your decisions and denote the calculation process.
A: Please see the explanation video that we recorded for this task: https://iris1.informatik.tu-chemnitz.de/index.php/s/dxg6yCy793ks2xi
Q: Algorithms such as Shannon-Fano, Huffman and Burrows-Wheeler Algorithm are described. Can you please tell which one is more important from an exam point of view? As one hour is not enough to solve all the algorithms.
A: They are all equally important. You should be able to solve all of them, but we will most probably only ask for one of them.
Q: In the Summer Semester 2019 exam, there is a question: Please write down a message with maximum and minimum entropy for the following character set: {0,1,2,3,4,5,6,7,8,9}. I understand that max entropy is when all characters are not the same in a message, so can we just answer (0,1,2,3,4,5,6,7,8,9) as maximum and (0,0,0,0,0,0,0,0,0) as minimum?
A: No, it's not that simple. If you have a look at the entropy formula, you'll find that the entropy of a message depends on the probability of occurrence of the individual characters. In the given task, the probabilities of occurrence of all characters are all the same, since each character occurs only exactly once in the alphabet (and since there is no other info on the statistics given). So a message with minimum entropy would be a message containing one character only, which could be any of the 10 characters, e.g. '5'. A message with maximum entropy here, would be an infinitely long string of any combination of characters from the alphabet, e.g. '123456789098765432123456.....'. This is because there is no maximum length of the message given. If a maximum message length would have been given in the task then the message with maximum entropy would be a string of that given maximum length, consisting of any combination of characters from the alphabet.
Q: Can you explain pattern substitution & LZW encoding?
A: Please be more specific in what exactly you don't understand about it. The consultation is not a repetition of entire tutorial tasks.
Q: Is the Burrows-Wheeler algorithm important from an exam point of view?
A: It is as important as the other algorithms.
Q: Can you plese explain LZW decoding?
A: Please see the solution slides of tutorial 3, slide 18 onwards.
Q: Pattern substitution: Zehn_zahme_Ziegen_zogen_zehn_zentner_Zucker_zum_Zoo
In the solution ehn_z and n_z are chosen. Can we take gen_z instead of n_z?
A: Yes, it is even more efficient. We have updated the solution slides accordingly.
Q: How many calculation and theory questions can we expect in the exam? Will it be of equal ratio like 7 theory and 7 calculation?
A: There are less calculation than there are theory questions. It's probably more like 1:3.