aelphy / isc Goto Github PK

Dear Mikhail,
I missed the seminar of the week and I am working on it now. I have two questions regarding it:

1. Do the points count as the grading of this seminar? If so can I get the points by submitting the seminar as an assignment?
2. Regarding the hyper-parameters optimization, do you plan to cover other approaches, e.g., Bayesian optimization, a sequential model-based optimization method for hyper-parameters optimization?

BR,
Caifa

week0:
Video intro about code execution (english): link dead

Hi Mikhail,

in notebook 5, what should be returned in the DFS and BFS algorithms?

I get the following error message from python:

  File "C:\APPS\Anaconda3_x86\lib\site-packages\IPython\core\interactiveshell.py", line 2961, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)


  File "<ipython-input-4-14cb4b2b62a6>", line 8, in <module>
    clock = get_ipython().run_line_magic('timeit', '- o fft(vector)')

  File "C:\APPS\Anaconda3_x86\lib\site-packages\IPython\core\interactiveshell.py", line 2131, in run_line_magic
    result = fn(*args,**kwargs)

  File "<decorator-gen-62>", line 2, in timeit

  File "C:\APPS\Anaconda3_x86\lib\site-packages\IPython\core\magic.py", line 187, in <lambda>
    call = lambda f, *a, **k: f(*a, **k)

  File "C:\APPS\Anaconda3_x86\lib\site-packages\IPython\core\magics\execution.py", line 1040, in timeit
    ast_stmt = self.shell.compile.ast_parse(transform(stmt))

  File "C:\APPS\Anaconda3_x86\lib\site-packages\IPython\core\compilerop.py", line 117, in ast_parse
    return self._fix_module_ds(compile(source, filename, symbol, self.flags | PyCF_ONLY_AST, 1))

  File "<unknown>", line 1
    - o fft(vector)

When I execute the second code cell.
The functions slow_fft(x) and fft(x) are implemented.

By the end of Problem1 in assigment1, you asked us to "Measure times of multiplications by T for different values of N and plot them in loglog scale."

I'm not sure what you want us to ask. By times of multiplications, do you want the number of multiplications of multiplying T with a vector, whose size is N?

Thank you very much!

Best,
Dai Qi

What is meant in homework by "You need to complete python intro notebook"? Should I clone the python_intro.ipynb file into my repository, work on it and then write a report about my findings according to everything what was meant to be done in homework 0?

What is the latest deadline for resubmissions of assignments?

I have 2 questions for the 4th assignment:

1. In the Intro to Numpy part there is stated: "For every problem provide examples of their usage." Does it pertain to the all the tasks from 1 to 14?

2. Mikhail, could you clarify what is meant by "Train your perceptron and report the results." in exercise Perceptron? Are we supposed to implement the multi-label classifier from scratch? Or can we use a ready implementation, such as sklearn.linear_model.Perceptron?

Hello, Mihail,
I am trying to add you as the collaborator on gitlab (The ISG gitlab). But I cannot find you. Can you just register on this Gitlab platform so that I can add you? Thanks.

What was this visualization workshop that we attended? Also, do you mean a "box and whisker" plot when mentioning the "mustache box" or span diagram?

Dear Mikhail,

I have check the content of assignment 2 and found its workload is too heavy.

For me, I have spent 4 hrs solving the first question (fft). Although the line of code is less than 20 lines, but it really cost a lot of time to fully understand the algorithm.

I also talked with one classmate and he said he spent 8 hrs but was still far from finishing it. While 8 hrs was already too much for a course with 3 credit points.

I hope you could reconsider the assignment and may possibly postpone the deadline. For me I think assignment2 deserves at least 3 weeks in total, just like a small project...

Thank you very much!

Best,
Dai Qi

Hi Mikhail!

For task 1, I wonder shall we calculate the method resolution order by programming? Or simply calculate manually and write down it in Markdown?

Best regards,
Dai Qi

The link provided about mapreduce no longer works. Digging around the website it appears that it is no longer available.
(http://kosmiktechnologies.com/blog/hadoop-mapreduce-training-hyderabad/)

Also, this project is designed for groups and is not clear what the 40 points is based on should we work on it as individuals. The exercise suggests groups of 3, each completing 3 functions from the set of tools. Does that mean we are expected to only complete 3 functions for the full project score as individuals?

Task 12 states that we have to "build a span diagram". What exactly is a span diagram?

Google did not help me in this.

"Page Not Found" for link https://berry.eecs.utk.edu/sc95/sc95.html in week2/Assignment2.ipynb - problem 5 - Folding-in technique

It appears there was no dictionary provided to map the highest ranking IDs to their corresponding name.

Any issues concerning installation can be posted here.

In this course, we'll use the following technology stack for numerical computation

• Git (version control)
• Docker (containers) only if you decide to work with it.
• Python + virtualenv / Anaconda
• jupyter (colab ?) notebooks
• numpy + scipy (matrices)
• scikit-learn (ml algorithms + pipelines)
• matplotlib (plotting)
• pandas (low scale data processing)
• probably? tensorflow and/or pytorch. (symbolic computation engine)

A simple roadmap to installing them can be found here -

• GIT
• Docker, Docker tutorial
• Python, Python tutorial + virtualenv, Anaconda, Anaconda tutorial, virtualenv tutorial
• numpy, scipy, scikit-learn, matplotlib, pandas are already included in anaconda distribution, you can install them separately with pip (see python + virtualenv tutorials)
• tensorflow, pytorch

The frameworks can be easily installed on Mac OS and Linux. Windows installation is, a bit tougher, so if you don't feel like it, try using docker (e.g. kitematic gui or console on windows).

If you run into any trouble, feel free to post here, even if it's like "i don't know what the hell all these letters mean!!!".

In warming up part of TensorFlow, where we are supposed to calculate the sum of squared numbers, the result for the code implemented in python is different than code implemented in Tensorflow (when calculating for big integers).

Up to 10**6, the result is the same, but with bigger numbers, the results do not match --> tensorflow returns number which is too small. As we discussed in a group of some students, is that an overflow error? Is there a way to allocate more memory per integer than int64 for tensorflow?

I am talking about that example:

# "i will insert N here later"
N = tf.placeholder('int64', name="input_to_your_function")

# a recipe on how to produce {sum of squares of arange of N} given N
result = tf.reduce_sum((tf.range(N)**2))