HoP Session Minutes

Notes taken by Wei Jin and Timothy on 23 Oct 2019.

Prior to the HoP session, HoP was provided with the consolidated feedback from students, collected by the ISTD Student Board.

Sections

Term 4 Subjects

Introduction to Algo

[Update] Course will have more advanced content added, with some simpler content being moved to Freshmore (Data-driven World).

[Feedback] Algo homework is too easy, the difficulty gap with the quiz is too big.

[Response] Will work with the lecturers mitigate the issue.

Computational Structures (CompStruct)

[Feedback] The workload is way too high for comp struct and it feels like the workload (such as 1D/Mojo) diverges from the learning objectives. Lucid has no proper documentation. Lucid is also not a transferrable skill.

[Response] Ideally it should be in Verilog, but that would take too long. The idea is to practice low level concepts like adders. Lucid is more convenient to finish the project within the allocated time. Constraint is important as it helps to develop skill sets that can be applied outside.

[Feedback] Why does the project have to be from a completely new concept when we already practice these concepts in lab. Could the project be an extension of the lab instead? There doesn’t need to be a physical project for it to be considered hands-on. After finishing comp struct, it feels like we still don’t understand any modern computer architecture. Doing the arcade machine (lot of time soldering and figuring out Lucid) is not worth the tradeoff not covering additional content about computer architecture. There can be coverage/overview on modern computer architecture.

[Feedback] The 2D tried to get us to optimise adders but they don’t give us any reference architecture. The results of the competition is also not shared after that, so other students cannot learn from the winning entries.

Term 5 Subjects

Elements of Software Construction (ESC)

[Feedback] The course does not meet its learning objectives and the project is not aligned with the things taught in class.

[Response] The course is going to be changed next Jan. We know that the project and whatever is taught is disconnected. Simon and Sudipta will revise the course and make some changes. Containers will also be included in the syllabus. The goal for the empty slot in Term 5 would be to also have another course covering software. Ideally the projects for Term 4 and 5 would be continuous, to build bigger software and see the full lifecycle of the project.

Term 6 Subjects

Database and Big Data

[Feedback] Too focused on AWS and website and not enough theory (lacks opportunity to practice these theories due to no homework), while specific platforms can be easily self-learnt. 30/45 marks just to set up infrastructure, last 15 marks doesn’t have anything to do with core objectives of database.

[Response] For the extra 15 marks, UI side might be out of scope of the course, but automation is an important part of databases. The course should give students a platform to think about the basics, but is not meant to spoonfeed students everything.

[Feedback] The rubrics are not scoped well to what is taught in class. Maybe the 30 basic marks can be further split into optimisation and be put into the project requirements since it is a learning objective. Makes it easier for students to argue for marks to negate bias and judgement.

[Response] If everything is too strict then everyone will have very similar projects. But ok.

[Feedback] The web dev aspect of the course is too heavy and out of scope of the course.

[Counter-Feedback] We have done web dev in ESC and it is not that difficult to accomplish.

[Feedback] Prof goes through theory but the project is more heavily focused on the practical aspects of AWS.

Computational Data Science (CDS)

[Feedback] Scope might be to broad, touch and go on a few topics.

[Response] Might have a data information retrieval or data-mining class to go more in depth.

[Feedback] The 3 hour lecture is too long. Can we split it into two?

[Response] The grouping of lessons is subject to everyone’s timetable. If the number of slots increases, then we would have more night courses which people would also complain.

Computer Engineering Track

[Response] There will be a class on the more CS-side of robotics, taught by Prof Malika. There will also be an IOT class to work on IOT platforms to reinvent the Computer Engineering track. The trend in the industry is that companies no longer outsource IOT projects due to security issues. Profs Binbin and Wenchao are involved in helping to plan the revised CE Track.

CS Foundation

[Feedback] There is a lack of Math foundation for people to understand computer science.

[Feedback] The current state of freshmore is too physics focused. It should be focused on algorithms instead because every pillar would need it. EPD-robotics, ESD-optimization, ASD-machine learning.

[Response] Some things are out of our control, some people feel that it is not necessary. Some people don’t want to do so much math. It is also difficult to put algo together with discrete math.

[Feedback] Explore the idea of half credits module like Bio/Chem in Term 1 to cover the other mods such as discrete math. Elective style of combining half modules.

[Response] Maybe can try out in Term 5, since there is a free slot for subject due to Prob&Stats being moved to Freshmore, but would need to see the response first. Maybe we can also allow Poly students to take Term 5 ahead of time as well (before Freshmore) if there are only two mods that term.

[Feedback] Proper introduction to C can also be included in the half modules. 2 hours is too little time for C, and 14 weeks might be overkill for one language.

[Response] Can maybe ask Simon to teach, because he taught C in Yale NUS.

[Feedback] It would be helpful if there was one pillar module before internship period.

[Response] Freshmore year is going to be changed to have two software related classes. We might also cover version control during one of the Freshmore mods.

Term 7/8 Issues

[Feedback] There is only one chance to take an elective in term 7/8 due to Capstone and HASS.

[Feedback] Half the courses in term 7/8 also require you to take machine learning in Term 6.

[Response] The job market requires a lot of machine learning application. Salary is higher in data analytics, so that is prioritised and those areas addressed first. We have not forgotten about the non-ML tracks.

Distributed Systems and Computing

The course focus is on GPGPU parallel programming, and then using the concepts learnt and applying to the design aspects of distributed systems. Will follow up with staff to make the course description clearer.

Staffing Updates

Incoming Profs