Feature
- Vocabulary: boost the representation power
- A Feature aggregation strategy is proposed to propagate information from early stags to the later ones. — (Li, et al. 2019) “Retinking on Multi-Stage Networks for Huma Pose Estimation”
- A multi-stage network is vulnerable by the information losing during repeated up and down sampling. To mitigate this issuage, a cross stage feature aggregation strategy is used to propagte multi-scale features form the early stages to the current stage in an efficient way.— (Li, et al. 2019) “Retinking on Multi-Stage Networks for Huma Pose Estimation”
- Features with different depth have different levels of abstraction of the image. — (Sun. et al. 2018) FishNet
- Feature concatenation is used when vertical and horizontal arrow meet. — (Sun. et al. 2018) FishNet
- In the context of the channels of a CNN, different channels are about differnt types of image features or raindrop features which cover a wide range of local image patterns. Deep Learning for Seeing Through Window With Raindrops
- Then feature seletection in a CNN is about assigning different weights to different feature channels. This motivates us to incorporate a channel attention/recalibration mechanism into the CNN. Deep Learning for Seeing Through Window With Raindrops
Cause and Effects
don’t use ‘Because’ as the first word in a sentence: it’s bad style.
There are a number of alternatives. You can use:
- ‘Due to…’; e.g. ‘Due to the heavy rain there was flooding.’
- ‘Owing to …’; e.g. ‘Owing to the heavy rain there was flooding.’
- ‘As…’; e.g. ‘As there was heavy rain, there was flooding.’
Gap
- Vocabulary: inconsequential, sufficiently, vulnerable, mitigate
- … is still in its infancy stage
- … contains noticeable artifacts.
Similarity
- Both … and … were fairly satisfied with…
- almost the same
- very similar, identical
- exactly, practically, approximately, almost, nearly, about, the same as
Novelty
- exploits …
- Note that the single stage module structure itself is not novel, but applying it in a multi-stage setting is new and shown effective in this work for the first time. — (Li, et al. 2019) “Retinking on Multi-Stage Networks for Huma Pose Estimation”
Improvement
- significiant, outstanding
- remarkable, obvious, notable (显著的) noticebale, brings noticebale performance improvement
Code will be released soon
- The source code of our algorithm will be made public soon to facilitate further research.
Writing Style
- https://elc.polyu.edu.hk/CILL/EAP/academicstyle.aspx
- http://www.uefap.com/writing/feature/intro.htm
- Avoiding the overuse and misuse of certain logical connectors, especially besides, furthermore and moreover. Besides is too informal, and both furthermore and moreover mean that the following information is more important than the information before, which is usually bad organisation. Use In addition or Also instead.
Describe the Figure
- See Fig. 1 for the illustration of real images taken through windows with raindrops
- the subjective is we, or the figure.
- the survey shows that….
- From/In the table, we can find/see that…
- It can be seen from the figure that…
- By comparing ## with ours, we can find that, ### can bring noticeable performance improvement, e.g. the PSNR improvement of ##dB is observed.
Comment
- Such results have demostrated the effectiveness of ###
- This demostrated that ###
To describe the trends
http://www.vocabulary.cl/ielts/academic-writing-task-1-describing-trends.htm
http://www.blairenglish.com/exercises/numbers_figures/numbers_selectionpage.html
https://www.ieltsbuddy.com/ielts-academic-writing-task-1.html
https://elc.polyu.edu.hk/CILL/exercises/cause_effect.aspx
Useful phrases for data description
- It can be seen from Table X that…
- As is shown in Figure Y, …
- As the findings in Figure Y reveal/show/indicate,…
- The effects of … on… are shown in Figure 3.
- According to Table Z, …
Some tips on verb tenses and voices
Content | Language Feature |
---|---|
Procedures of the study | Passive Voice and Past Tense |
The instrument itself | Present Tense or Past Tense |
Figure
- The network is exemplified in Figure. — (Li, et al. 2019) “Retinking on Multi-Stage Networks for Huma Pose Estimation”
Conclusion
The Conclusion chapter usually contains five elements in the following order:
-
Restatement of the study’s purpose
Recapitulate the main points by stating again
(i) the situation in the field before your study was carried out, i.e. the problems and gaps;
(ii) the study’s purpose; and
(iii) your research questions (and hypotheses, if any). -
Summary of the major findings
Present succinctly your answers to each of your research questions.
State your proven hypotheses, if any. -
Significance and implications of your study
Stress how your findings have improved the situation by addressing the gaps and dealing with the problems.
Make it clear to what extent you have filled the gaps or alleviated the problems.
Refer to previous research for comparison or support.
Explain any unexpected outcomes and results which contradict previous findings.
Articulate the contributions of your findings (and/or methodology) to your field.
Discuss the applicability of your methodology and findings to other contexts, and the generalisability of your results.
Do not over exaggerate your achievements. -
Limitations of your study
Be critical of your study.
State the weaknesses of all aspects of your study, e.g. methodology and generalisability.
Acknowledge any mistakes made, and the problems and weaknesses that remain.
Show that you are a reflective researcher. It is better that you point out your limitations than have the examiners point them out for you. -
Recommendations for future research
State the need for further research.
Make concrete and constructive suggestions about future lines of investigation.
Provide brief justifications.