Source author record

Wouter Kuijper

Wouter Kuijper appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

Data Structures and Algorithms Machine Learning Neural and Evolutionary Computing Programming Languages

Catalog footprint

What is connected

3works

4topics

3close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2016arXiv

Geometric Memory Management

In this report we discuss the concepts of geometric memory align- ment, geometric memory allocation and geometric memory mapping. We introduce block trees as an efficient data structure for representing geo- metrically aligned block allocation states. We introduce niche maps as an efficient means to find the right place to allocate a chunk of a given size whilst maintaining good packing and avoiding fragmentation. We intro- duce ledging as a process to deal with chunk sizes that are not a power of two. We discuss using block trees for memory mapping in the context of virtual memory management. We show how ledging can also be applied at the level of virtual memory in order to create fixed size virtual mem- ory spaces. Finally we discuss implementation of both geometric memory allocation as well as geometric memory mapping in hardware.

preprint2016arXiv

The Artificial Mind's Eye: Resisting Adversarials for Convolutional Neural Networks using Internal Projection

We introduce a novel artificial neural network architecture that integrates robustness to adversarial input in the network structure. The main idea of our approach is to force the network to make predictions on what the given instance of the class under consideration would look like and subsequently test those predictions. By forcing the network to redraw the relevant parts of the image and subsequently comparing this new image to the original, we are having the network give a "proof" of the presence of the object.

preprint2011arXiv

Generic Programming of Reusable, High Performance Container Types using Automatic Type Hierarchy Inference and Bidirectional Antichain Typing

We introduce a new compile-time notion of type subsumption based on type simulation. We show how to apply this static subsumption relation to support a more intuitive, object oriented approach to generic programming of reusable, high performance container types. As a first step towards an efficient implementation of the resulting type system in a compiler we present a novel algorithm for bidirectional type inference over arbitrary syntax graphs. The algorithm uses the new static type subsumption relation to compress the data that has to be stored for each node in the typeflow graph. During typeflow analysis this means that the set of types for a given node can be symbolically represented using antichains instead of using bitvectors or some other explicit set representation. This results in a typing algorithm that is both flexible and precise and shows good performance on representative instances.