15 problems
- 0 votes0 replies1 view
Bubeck–Li–Nagaraj robustness law for arbitrary-weight two-layer networks
Bubeck–Li–Nagaraj conjecture. The stated lower bound should hold for every Lipschitz activation and arbitrary weights, with no restriction on their magnitudes. The conjecture quant…
- 0 votes0 replies0 views
Conjecture on non-robust useful features limiting robustness accuracy
Let be a feature associated with an operator acting on quantum states, let be a distribution of labelled states…
- 0 votes0 replies0 views
Structural-property conjecture for randomly sampled dynamical models
Consider a family of dynamical systems sharing an interconnection structure and a property of interest. The property is structural when it holds for the whole family,…
- 0 votes0 replies0 views
Extension of robust Robinson-property recovery to -graphons for
Let be an -graphon, and let the results of Theorem cited in the source be understood as the robust Robinson-property recovery results established there for . Robust r…
- 0 votes0 replies0 views
CODASCA's lack of robustness to corrupted labels in federated deep AUC maximization
Let CODASCA denote the min-max federated deep AUC maximization algorithm, and suppose it uses the square loss, which is not symmetric: a loss is symmetric when i…
- 0 votes0 replies0 views
Conjecture on the maximal insertion model withstandable by the coloring algorithm
Insertion-model conjecture. The most general insertion model that the algorithm can withstand is one in which the original color classes are respected.
- 0 votes0 replies1 view
Losslessness conjecture for dissipativity-based robustness analysis with terminal-cost IQCs
Consider robustness analysis using the dissipativity-based results of Theorems and, and compare it with robustness analysis using Theorem with general O'Shea-Zames-Falb multipliers…
- 0 votes0 replies1 view
Conjecture on δ-robustness for k-nearest-neighbor graph weights
Let be the domain, let be the data set of points, and let be the number of nearest neighbors. Set … Let be defined by the Laplace-learning strategy in th…
- 0 votes0 replies0 views
Non-robustness of the Bid Adjustment Algorithm against weak incentives to deviate
Let a generator's weak incentive to deviate mean that there exists an execution of the deviation dynamics along which … so that the generator obtains a higher payoff than…
- 0 votes0 replies0 views
Robust extinction conjecture for theta-perturbations
Let denote the stochastic population process, and let a -perturbation of be the process defined by equation. In the setting where and…
- 0 votes0 replies1 view
Conjecture on robustness results for integrable Nemitski losses
A pairwise loss function is a function on pairs of examples and prediction values, and its associated risk functional is obtained by integrating the loss with respect to the releva…
- 0 votes0 replies0 views
Exponential decay conjecture for the robustness parameter of minimally redundant frames
Let be a frame with for all , and let denote the robustness parameter discussed above. Exponential de…
- 0 votes0 replies1 view
Monotonicity of robustness across equilibrium classes
Let denote the maximal robustness of the three-dimensional equilibrium class indexed by . Equilibrium-class monotonicity conjecture. If and ,…
- 0 votes0 replies0 views
The robustness equality for the equilibrium classes and
For the three-dimensional equilibrium classes indexed by pairs and , let and denote their maximal full robustness, and let denote…
- 0 votes0 replies0 views
The planar robustness conjecture for regular polygons
For , let denote the maximal robustness among planar convex solids in the equilibrium class with points, and let denote the robustnes…