-
**Background:** The existing `ssautil.AllFunctions` helper performs a reachability analysis starting from all the packages in a Go SSA program, and returns the set of functions it encounters:
```go…
-
We should try a simple version of function return approximations. However this should probably not happen before 4.03 unless we run out of other things to do.
-
Given that `ReLU(x) = x (0.5 + 0.5 sgn(x))`, this reduces to approximating the sign function, and this paper appears to have the state of the art: https://eprint.iacr.org/2020/834
Also note
- `…
j2kun updated
2 months ago
-
In Laplace approximation, the Hessian of the loss function is computed for quadratic approximation. Can this package be used to do a block-diagonal approximation of the Hessian at the minimum? If yes,…
-
Dear All,
in trying to learn a little bit about neural networks I came across tiny-dnn,
which appears to be nicely suited for someone familiar with C++.
As an initial test I tried to construct a ne…
-
This should include at least:
- [x] Chebyshev interpolation point computation
- [ ] Chebyshev polynomial computation
- [x] Lagrange interpolant computation
- [ ] Newton interpolation
- [ ] Herm…
-
MathOptInterface includes a function `eval_constraint_jacobian_transpose_product` which some solvers, like MadNLP) can use to avoid materializing the constraint Jacobian. MadNLP will error when using …
-
Table of values such as probability of relative spawning biomass falling below overfishing threshold in 2019, which I think would be calculated from "model" as
pnorm(model$minbthresh, model$derived_…
-
This is implemented by `MiniMaxApproximation` in Mathematica, see:
http://reference.wolfram.com/mathematica/FunctionApproximations/tutorial/FunctionApproximations.html
http://reference.wolfram.com/ma…
-
When I use the pywt.swt function for a level 1 transform I get different results than when doing a manual convolution with the low pass and high pass filters using wavelet.dec_lo and wavelet.dec_hi. I…