Problem Selection - KTH

Lecture 2: Problem Selection

Overview • “How to have a bad career in research/ academia”, the talk by David Patterson at UC Berkeley, 1994 (shortened to adopt for PhDs) • Analysis of Assignments 1 • Next assignment

p. 2 - FIL3001- The Art of Doctoral Research

How to have a bad career in research • Part I: Key Advice for a Bad Career • Part II: Key Advice on Alternatives to a Bad Career • Topics covered in both parts • • • • • •

Selecting a Problem Picking a Solution Performing the Research Evaluating the Results Communicating Results Transferring Technology p. 3 - FIL3001- The Art of Doctoral Research

Bad Move #1: Be THE leading expert • Invent a new field! • Make sure its slightly different

• Be the real Lone Ranger: Don’t work with others • No ambiguity in credit • Adopt the Prima Donna personality

• Research Horizons • • • •

Never define success Avoid Payoffs of less than 20 years Stick to one topic for whole career Even if technology appears to leave you behind, stand by your problem p. 4 - FIL3001- The Art of Doctoral Research

Bad Move #2: Let Complexity Be Your Guide • Best compliment: “Its so complicated, I can’t understand a thing you’ve said!”

• Easier to claim credit for subsequent good ideas • If no one understands, how can they contradict your claim?

p. 5 - FIL3001- The Art of Doctoral Research

Bad Move #3: Never be Proven Wrong • Avoid Implementing • Avoid Quantitative Experiments – If you’ve got good intuition, who needs experiments? – Why give grist for critics’ mill? – Takes too long to measure

• Avoid Benchmarks • Projects whose payoff is ≥ 20 years gives you 19 safe years

p. 6 - FIL3001- The Art of Doctoral Research

Bad Move #4: Use the Computer Scientific Method Obsolete Scientific Method:

Computer Scientific Method:

• • • •

• Hunch • 1 experiment & change all parameters • Discard if doesn’t support hunch • Why waste time? We know this

Hypothesis Sequence of experiments Change 1 parameter/exp. Prove/Disprove Hypothesis • Document for others to reproduce results

p. 7 - FIL3001- The Art of Doctoral Research

Bad Move #5: Avoid Feedback • Always dominate conversations: Silence is ignorance – Corollary: Loud is smart

• Reviews – If it's simple and obvious in retrospect => Reject – Quantitative results don't matter if they just show you what you already know => Reject – Everything else => Reject

• Don’t read • Don’t be tainted by interaction with users or industry p. 8 - FIL3001- The Art of Doctoral Research

Bad Move #6: Publishing Journal Papers IS Technology Transfer • Target Archival Journals: the Coin of the Academic Realm – It takes 2 to 3 years from submission to publication => timeless

• As the leading scientist, your job is to publish in journals; it’s not your job to make you the ideas understandable by an ordinary engineer • Going to conferences and visiting companies just uses up valuable research time p. 9 - FIL3001- The Art of Doctoral Research

Bad Move #7: Writing Tactics for a Bad Career • Papers: It’s Quantity, not Quality – Personal Success = Length of Publication List – “The Least Publishable Unit” is Good Enough for You 1 idea 4 journal papers 16 extended abstracts 64 technical reports

• Student productivity = number of papers – Number of students: big is beautiful – Never ask students to implement: reduces papers

• Legally change your name to Aaaanderson p. 10 - FIL3001- The Art of Doctoral Research

5 Writing Commandments for a Bad Career 1. 2. 3. 4. 5.

Do not define terms, nor explain anything Publish before implementing Replace “will do” with “have done” Do not mention drawbacks to your approach Do not reference any papers

p. 11 - FIL3001- The Art of Doctoral Research

6 Talk Commandments for a Bad Career 1. 2. 3. 4. 5. 6.

Do not illustrate Do not welcome brevity Do not print large Do not use colors Do not skip slides in a long talk Do not practice

p. 12 - FIL3001- The Art of Doctoral Research

Summary of Part I • Part I: Key Advice for a Bad Career: – – – – – – –

Invent a field and stick to it Let complexity be your guide (confuse enemies) Never be proven wrong Use the computer scientific method Avoid feedback Publishing journal papers is technology transfer Write many (bad) papers by following 5 writing commandments – Give bad talks by following 6 talk commandments p. 13 - FIL3001- The Art of Doctoral Research

Part II: Key Advice on Alternatives to a Bad Career • Goal is to have impact • 4 Steps 1. 2. 3. 4.

Selecting a problem Picking a solution Quantitative evaluation Transferring technology

p. 14 - FIL3001- The Art of Doctoral Research

Selecting a Problem Invent a new field & stick to it? • No! Do “Real Stuff”: solve problem that someone cares about

p. 15 - FIL3001- The Art of Doctoral Research

Picking a solution Let Complexity Be Your Guide? • No! Keep things simple unless a very good reason not to – Best results are obvious in retrospect: “Anyone could have thought of that”

Use the Computer Scientific Method? – No! Run experiments to discover real problems – Use intuition to ask questions, not answer them

p. 16 - FIL3001- The Art of Doctoral Research

Evaluating Quantitatively Never be Proven Wrong? • No! If you can’t be proven wrong, then you can’t prove you’re right – “Better to curse the candle than curse the darkness.”

• Report in sufficient detail for others to reproduce results – Can’t convince others if they can’t get same results

p. 17 - FIL3001- The Art of Doctoral Research

Transferring technology (by convincing others) Publishing Journal Papers IS Technology Transfer? • No! Selecting problem is key: “Real stuff” – Ideally, more interest as time passes

• Change minds with believable experiments

p. 18 - FIL3001- The Art of Doctoral Research

Summary of Part II • Goal is to have impact • Feedback is key: Seek out & value critics • Do “Real Stuff”: Make sure you are solving a problem that someone cares about • Taste is critical in selecting research problems, solutions, experiments, and communicating results; acquired by feedback

p. 19 - FIL3001- The Art of Doctoral Research

Analysis of Assignment 1 • The goal is to give your critical opinion in order to help the author to see the weak points • If you are the author – don’t answer to criticism, we want to keep the discussion anonymous • Don’t get upset it your work is criticized – this is how we learn - from our mistakes (and mistakes of others)

p. 20 - FIL3001- The Art of Doctoral Research

Goal • Goal should answer the question “What do you want to achieve”? – Typically the goal is to solve some problem – A well-formulated goal makes clear which problem is addressed

p. 21 - FIL3001- The Art of Doctoral Research

Examples of goals • Energy Efficient Machine-Type Communications (MTC) over Cellular Networks • A formal model for describing and verifying the operational semantics of Big Data Analysis Systems • Low power neural recording interface • RF Devices and Circuits For Extreme Temperatures

p. 22 - FIL3001- The Art of Doctoral Research

Examples of goals, cont. • Enhance security of big data in distributed environments • Efficient monitoring for edge cloud computing • Design methodologies for monlithic 3D Ge nanowire circuits • Efficient and secure wireless caching for 5G RAN • Reveal ultrafast dynamics in metalorganic halide perovskite material and solar cell device

p. 23 - FIL3001- The Art of Doctoral Research

Hints for formulating goals • Use one action verb • Try to show purpose and contribution, if possible • Decribe application area, if possible

p. 24 - FIL3001- The Art of Doctoral Research

Objectives • Objectives provide a more detailed breakthrough of the goal – They make more clear which benefits will be received by the target group

– Because a goal is at a high-level, it may take more than one project to achieve it. However, an objective should be achievable.

p. 25 - FIL3001- The Art of Doctoral Research

Energy Efficient Machine-Type Communications (MTC) over Cellular Networks • To evaluate performance of existing MAC for MTC and propose an improved MAC • To develop and implement techniques for minimizing losses within the converter and power consumption of the control

p. 26 - FIL3001- The Art of Doctoral Research

A formal model for describing and verifying the operational semantics of Big Data Analysis Systems • Define what Big Data Analysis System means • Compare language expressiveness of different Big Data Systems • Verification of provenance in analytical data products generated within Big Data Analysis Systems • To develop a type theoretic understanding of machine learning tasks and optimization tasks

p. 27 - FIL3001- The Art of Doctoral Research

Low power neural recording interface • System level implementation of a novel architecture in MATLAB/Simulink • Transistor level implementation of the system and mask design in a 180 nm technology • To estimate the influence of substrate noise • Measurement of the chip performance

p. 28 - FIL3001- The Art of Doctoral Research

Results and Indicators • Results describe the services or products to be delivered to the intended target group – It should be possible to measure results through the use of objective indicators – It is very important to know who is your target group • For whom your problem is a real problem?

• Indicators are used to evaluate the success

p. 29 - FIL3001- The Art of Doctoral Research

Examples of results and indicators • Objective: Energy Efficient Machine-Type Communications (MTC) over Cellular Networks • Result: A new MAC scheme for cellular MTC • Indicator of success: Serving 1,000,000 devices per 𝑘𝑚2, and assuring over 10 years battery lifetime for them

p. 30 - FIL3001- The Art of Doctoral Research

Examples of results and indicators Objective: A formal model for describing and verifying the operational semantics of Big Data Analysis Systems • Result: A semantics to model data dependencies between analytical products and data sets • Indicator of success: (1) Enables tracing and invalidation of data products, (2) Leads to an operational semantics for analytical workflows

p. 31 - FIL3001- The Art of Doctoral Research

Examples of results and indicators • Objective: Low power neural recording interface • Result: Analysis of the influence generators on the system performance • Indicator of success: Chip prototype working according to the specifications

p. 32 - FIL3001- The Art of Doctoral Research

My wishes for your next assignments • Start working on the assignment earlier than the deadline day – Good ideas need time to ”ripe”

p. 33 - FIL3001- The Art of Doctoral Research

Second Assignment • Write a ½ page text which answers the following question related to the problem which you have targeted to solve in your 1st assignment:

If this problem is so important, why no one have solved it so far?

p. 34 - FIL3001- The Art of Doctoral Research

Second assignment, cont. • The assignment is due on Sunday, Oct 16th. Send it to me by email ([email protected]) as a pdf file. • Please call the attached file according to the following convention: • FirstnameFamilyname2.pdf

• For example, the file of my attachment would be called ElenaDubrova2.pdf • Use the same convention for other assignments • Put in the subject line: FIL3001 Assignment 2 p. 35 - FIL3001- The Art of Doctoral Research

Hints for Assignment 2 A problem may still be unsolved because: 1. It only occured recently due to some changes in technology or new discoveries – Complex problem are usually solved in a ”pyramid” way

2. It is an old problem, and many people tried to solve it, but noone have succeeded so far. But: – Some new facts have been discovered recently which can assist you in solving it – The problem requires some unique combination of expertise which very few people have p. 36 - FIL3001- The Art of Doctoral Research

Other reasons … 3. Previously proposed solutions were OK for the old conditions, but are not OK in the changed ones 4. A problem may also be unsolved because: – It is too hard – It is not important – The topic is out of fasion

5. Or, may be the the problem is solved already, but your don’ know about it

p. 37 - FIL3001- The Art of Doctoral Research