rahulnair23

5 ms·

1.
by rahulnair23 7mo ago
Years ago, we build systems to predict delays on the German rail network. Paper here: https://www.sciencedirect.com/science/article/pii/S0968090X1... The complexity of operations is astounding and the organis
2.
by rahulnair23 1y ago
Isolating tech goes earlier than that. I know all my neighbours in my block who walk, cycle, or take the bus. Repeated interactions. The ones who take the elevator to the basement car park are the ones I never cross paths with.
3.
by rahulnair23 1y ago
Interesting work. For a slightly different take using a similar intuition, see our paper [at ACL 2024]( https://arxiv.org/abs/2402.14860 ) on ranking LLMs which may be of interest. Our HuggingFace space has some examples
4.
by rahulnair23 2y ago
Pointers to said tools would be beneficial. Robust optimization is somewhat different from what the author states.
5.
by rahulnair23 2y ago
The ISO link[1] says this standard has been withdrawn however. [1] https://www.iso.org/obp/ui/#iso:std:iso:3103:ed-1:v1:en
6.
Ranking Large Language Models Without Ground Truth(arxiv.org)
2 points by rahulnair23 2y ago|0 comments
7.
by rahulnair23 3y ago
Washington D.C. is a great example of this. While looking at residential location choice, around 10 years ago, you could discern a boundary (Georgia Ave in those days). Netflix used to show the most popular movies by zip code, and one side
8.
by rahulnair23 4y ago
Always jarring to hear the claim > Technology is fundamentally neutral It is in the same way guns are fundamentally neutral. You can't view it without context. Include that and it is clear that tech (or guns) isn't neutral at a
9.
by rahulnair23 5y ago
Yes, Corbusier's vision of cities was tied to speed, and by consequence, to cars. But Chandigarh today has fantastic segregated bike infrastructure! No other Indian city comes close. The city itself is pan-flat and relatively small (sh
10.
by rahulnair23 5y ago
Many causes have been put forward for this. However, I'm not sure the all safety aspects here have been well understood. Here in Dublin, Ireland the few weeks immediately after the lockdowns where horrendous for driver behaviour and ag
11.
by rahulnair23 5y ago
One way to tell if a Machine Learning model is any good is to see how it does on unseen/new patients. Of course, we don't wait to try it on real patients, so typically you'd partition the data you have already into (a) what y
12.
by rahulnair23 5y ago
From the full paper[1]: > All models were trained on 70% of the data and tested on the remaining 30% of the data. Note that each month record for each participant was considered an independent data point. I'm almost certain that thi
13.
by rahulnair23 6y ago
I have not stated "all screening is bad". Broad-based population screenings as the parent comment suggests, in my opinion, are. I'm yet to see any clinically-valid distinguishing aspects that would suggest AI would add value
14.
by rahulnair23 6y ago
Health policy is fraught with counter-intuitive phenomenon - and screening is one of them. Seems like it should help, but in practice leads to over-diagnosis. For example - Cancer rates jumped in Korea after screening with no impact on pati
15.
by rahulnair23 6y ago
One place this is abundantly clear is in the humanitarian sector, which I peeked into for a couple of years for a project. For all the talk about participatory processes, the entire system is donor-focused. Its an army of well meaning folks
16.
by rahulnair23 6y ago
Is there an accepted threshold for tropospheric NO2 concentrations above which it is considered harmful?
17.
by rahulnair23 6y ago
Inequity, education, sustainability, public policy, governance - few examples where techno-solutionism isn't the right approach.
18.
by rahulnair23 7y ago
Here is a revised link [1] to the full report. Table 1, on Page 10 shows a summary. [1] https://dms.ntsb.gov/pubdms/search/document.cfm?docID=477717...
19.
by rahulnair23 7y ago
I'd suggest reading the NTSB report on the pedestrian killed by Uber's "self-driving" vehicle [0]. The vehicle misclassified Elaine Herzberg so many times in the seconds leading up to taking her out at 40 mph - its clear
20.
by rahulnair23 7y ago
I think you've missed the point, going by your second paragraph. How capacity is provisioned from A to B influences how people choose to move and, over the long term, build homes/workplaces. Induced demand does not suggest that su
21.
by rahulnair23 7y ago
Fully agree on the economic merits of English here. I think the point is more than methodological though. It is one of equity and access. Broad swaths of society don't have the same access to the technology. My mother tongue Malayalam
22.
by rahulnair23 7y ago
This is absolutely spot on. A good deal of resources in some large organizations are wasted on vanity projects initiated by mid-management specifically for such value-capture.
23.
by rahulnair23 7y ago
Developed and developing are also terms that are arbitrary. Human Development Indices of some countries that are considered developing today exceed those of developed countries when these terms were first used. In common use it implies a st
24.
by rahulnair23 8y ago
Governments do fall under the purview of GDPR. Do you have a source that states otherwise?
25.
by rahulnair23 9y ago
Really wonderful interview. One point that doesn't come up os that serious operational inefficiencies and "friction" exist in how current fixed route services are delivered in most cities. Some due to political challenges (se
26.
by rahulnair23 10y ago
Grad students are absolutely that. During my grad student days in a STEM program at a large state university, professors were occasionally able to "buy-out" of their regular course load and forfeit a percentage of their salary. Th
27.
by rahulnair23 12y ago
Fascinating and insightful article. I would still consider some cases where information is useful. (a) There are some really interesting community-based initiatives where urban dashboards add value albeit still data-driven. Instead of using
28.
by rahulnair23 13y ago
SuperShuttle is an interesting service. I used it a fair bit in the DC metro area, since Dulles Airport is not very accessible by public transport. My main observation was that your circuity factor (the time deviation from direct travel tim
29.
by rahulnair23 13y ago
Yes travel time includes time to walk from the Cell tower (considered the "source" and "sink") to the stops. We could generate passenger level stats for the sample we observed, but have presented it at the route level for now, since its tar
30.
by rahulnair23 13y ago
Hi, I'm from the IBM Research team who worked on this. The reduction is in total system-wide journey times in passenger-minutes (travel time + expected waiting time). The model makes underlying assumptions on how people arrive at stops to

More ›