Konstantinos Samaras-Tsakiris

Compute supertype parameters

3

I experimented with the following behavior, which is not supported by this package: ``` struct CustomArray{N1,N2}

More Transformer architectures

3

Hi! We're implementing some transformer architectures like [Gated Transformer Networks for Multivariate Time Series Classification](https://arxiv.org/abs/2103.14438). Are you interested in pulling more examples like this?

Tracking on Corner-Poor Environments: Alternate Feature Detectors/Descriptors

4

Hello, It would be interesting to combine this SLAM system with a feature detector and descriptor _other than **ORB**_, primarily to compare performance in special use cases, such as _environments...

fix(generator): don't truncate datetime field precision

2

closes #804 - [ ] Unit tests for the changes exist - [ ] Tests pass without significant drop in coverage - [ ] Documentation reflects changes where applicable -...

Datetime microsecond precision dropped by query builder

6

## Bug description I have a model with a column of type `DateTime @db.timestamptz(6)` as primary key. Microsecond precision is essential in this application. I try to create multiple rows...

Regions are a more intuitive concept to use HTM, exemplified in the test [test_projection.jl](test/test_projection.jl). Needs docs.

Rethink SP local inhibition

1

The SP tiebreaker is necessary to enforce strict sparsity control, but it has undesirable sideffects in other properties: - `sp(z)` becomes stochastic (random seed from hash(z)?) - noise robustness suffers...

Numenta's latest paper [Grid Cell Path Integration For Movement-Based Visual Object Recognition](https://arxiv.org/abs/2102.09076) implements an encoder for MNIST. In the paper they use a CNN with a k-winner layer. I wonder...

MLJ integration

1

relevant discussions: - https://github.com/alan-turing-institute/MLJ.jl/issues/60#issuecomment-572517203 - [MLJTime](https://github.com/alan-turing-institute/MLJTime.jl) & https://github.com/alan-turing-institute/MLJ.jl/issues/303 - https://github.com/alan-turing-institute/MLJ.jl/issues/51 Some of these things need to be developed on MLJ's side first.

Explore SparseArrayKit.jl (DOK format, N-D tensors)

1

https://github.com/Jutho/SparseArrayKit.jl ## N-D tensors Synapses are by nature 2D (pre-, post-synaptic addresses). In case of multiple spatial dimensions, no reason to burden the synapse matrices with those but rather should...