I feel that #Geospatial Data Industry has been a bit stagnant in terms of true #innovation. There is a lot of exciting work continue to happen in this space. I had a go at exploring how Vector Embeddings will revolutionise the Geospatial Data Industry. https://t.co/xLezwC7oiO
@helenmakesmaps A very good blog π
In a training sess, many moons ago, we discussed exactly the same issue. I am a big fan of humility, so the title of this book by @stefaniesw was very challenging to accept. It still is. But it has some great tips.
F*ck Being Humble https://t.co/iDkJDIsqTa
I have always been a huge fan of Umbra as a leading EO organisation. This is nothing but dominating thought leadership, in the world where data hoarding is a disease. Well done @umbraspace and @mouthofmorrison take a bow πββοΈ
Umbra shares all data under the Creative Commons license (CC BY 4.0), granting you the freedom to publish images and link the underlying data as needed. The license requires you to credit us ('attribution'), but no need to ever ask us for permission to use.
@mouthofmorrison Just a hunch aka gut feeling, but we are on the cusp of exploiting the TRUE potential of EO data; something more than base layers. EO data will & has already started to fuel the analytics of the most powerful systems. How Bloomberg data powered the fintech revolution. Bring it on
Why are square matrices particularly interesting?
Square matrices represent linear maps from a space to itself, which allows for meaningful comparisons between inputs and outputs since they belong to the same space.
For instance, if π:ββ΄βββ΄ it makes sense to ask when π(a) is parallel to a, since π(a) and a lie in the same space. However, asking when π(a) is parallel to a for π:ββ΄ββΒ³ doesn't make any sense, since π(a) and a are different types of objects.
Another example is the determinant. In square matrices, determinants measure how a linear map alters (either expanding or shrinking) a unit of volume. For instance, the determinant of the transformation (x,y,z)β¦(-2x,2y,2z) is -8, indicating an 8-fold increase in volume but with a reversed orientation.
If we try to go from 3D to 2D (2x3 matrix) and use the same idea: how much area does a given volume ends up producing? we run into problems:
- When transitioning from 3D to 2D, the "stretching factor" is not consistent. Take the projection map (π₯,π¦,π§)β¦(π₯,π¦), as an example and consider the effects when elongating a volume vertically.
- Transitioning from 2D to 3D invariably results in no volume since the starting dimension is inherently insufficient. Thus, regardless of the transformation, our "stretching factor" invariably remains 0.
Simply put, when dealing with non-square matrices, the concept of the "determinant" can be either poorly defined or trivially zero due to oversimplified reasoning.
Back on the saddle, pedaling 160km for @Shelter again, with less training this year but your support is overwhelming! Close to my target in 24hrs. Please donate as much from Β£1 to a million. I'll power my ride & provides shelter. 2 weeks to go! π΄π π
https://t.co/I5NIpGQ1wq
@KAnderson_RS Congratulations Karen. This is awesome. The impact and influence that you bring to the industry is invaluable, both academia and industry. Really proud to call you a friend. π
@amirhusain_tx Coffee is my poison (4-5 double espresso), then it's green tea (3 odd). I then consume around 1.5L of water in the evening. More water than hot beverages and plenty of visits to the loo π . My coffee is 36g each shot, around 200ml per glass of tea.
What have you made me do π«£
1/Several leaders share their hopes for AI in 2023, including finding key missing pieces that will enable algorithms to reason, building a personal data timeline, improving AI processes, discovering new principles for explainability, and using generative AI for active learning.