Classiication is a fundamental act of cognition. It underlies nearly all learning, transfer of learning, generalization and abstraction. One of the well-known hard and fundamental problems is the one of perceptual aliasing, i.e. that the sensory stimulation caused by one and the same object varies enormously depending on the distance from the object, orientation, lighting conditions, etc. In th...

Physically Unclonable Functions (PUFs) have been an emerging topic in hardware security and trust in recent years, and many different kinds of PUFs have been presented in the literature. An important criterion is always the diversity of PUF responses for different devices, called inter-device uniqueness. A very popular uniqueness metric consists of calculating the pairwise hamming distance betw...

In this paper we show how to carry out robust place recognition using both near and far information provided by a stereo camera. Visual appearance is known to be very useful in place recognition tasks. In recent years, it has been shown that taking geometric information also into account further improves system robustness. Stereo visual systems provide 3D information and texture of nearby regio...

We target the problem of closed-loop learning of control policies that map visual percepts to continuous actions. Our algorithm, called Reinforcement Learning of Joint Classes (RLJC), adaptively discretizes the joint space of visual percepts and continuous actions. In a sequence of attempts to remove perceptual aliasing, it incrementally builds a decision tree that applies tests either in the i...

Antialiasing is a necessary component of any high quality renderer. An antialiased image is produced by convolving the scene with an antialiasing filter and sampling the result, or equivalently by solving the antialiasing integral at each pixel. Though methods for analytically computing this integral exist, they require the continuous two-dimensional result of visible-surface computations. Beca...

Many algorithms exist which generate per-pixel geometry by selectively discarding fragments generated for a simple bounding geometry. On the other hand, multisampling support has become ubiquitous and is almost free in current graphics hardware. In this paper we leverage the ability of setting a pixel’s coverage mask in the pixel shader to make seemingly inherently pixel-based per-pixel geometr...

A parametric model and a corresponding parameter estimation algorithm for unwrapping two-dimensional phase functions, are presented. The proposed algorithm performs global analysis of the observed signal. Since this analysis is based on parametric model tting, the proposed phase unwrapping algorithm has low sensitivity to phase aliasing due to low sampling rates and noise, as well as to local e...

Affine type systems manage resources by preventing some values from being used more than once. This offers expressiveness and performance benefits, but difficulty arises in interacting with components written in a conventional language whose type system provides no way to maintain the affine type system’s aliasing invariants. We propose and implement a technique that uses behavioral contracts t...

Reinforcement learning is an ideal framework to learn visual routines since the routines are made up of sequences of actions. However, such algorithms must be able to handle the hidden state (perceptual aliasing) that results from visual routine’s purposefully narrowed attention. The U-Tree algorithm successfully learns visual routines for a complex driving task in which the agent makes eye mov...

Aliasing refers to the phenomenon that high frequency signals degenerate into completely different ones after sampling. It arises as a problem in context of deep learning downsampling layers are widely adopted architectures reduce parameters and computation. The standard solution is apply low-pass filter (e.g., Gaussian blur) before (Zhang in: ICML, 2020). However, it can be suboptimal same acr...