Koopman NMPC Koopman based Learning and Nonlinear Model Predictive Control of Control affine Systems

Using the Koopman canonical transform, control-affine dynamics can be expressed by a lifted bilinear model . The model is used for nonlinear model predictive control (NMPC) design . The benefits are highlighted through an example of a simulated planarquadrotor . Prediction error is greatly reduced and closed loop performancesimilar to NMPC with full model knowledge is achieved .…

MUSER MUltimodal Stress Detection using Emotion Recognition as an Auxiliary Task

The capability to automatically detect human stress can benefit artificialintelligent agents involved in affective computing and human-computer interaction . Stress and emotion are both human affective states, and stress has important implications on the regulation and expression of emotion . We propose MUSER — a transformer-based model architecture and a novelmulti-task learning algorithm .…

StrobeNet Category Level Multiview Reconstruction of Articulated Objects

We present StrobeNet, a method for category-level 3D reconstruction of objects from unposed RGB images . Reconstructing general articulating object categories has important applications, but is challenging since objects can have wide variation in shape, articulation, appearance and topology . We address this by building on the idea ofcategory-level articulation canonicalization — mapping observations to acanonical articulation which enables correspondence-free multiview aggregation .…

Sound Event Detection with Adaptive Frequency Selection

In this work, we present HIDACT, a novel network architecture for adaptivecomputation for efficiently recognizing acoustic events . We evaluate the model on a sound event detection task where we train it to adaptively processfrequency bands . The model learns to adapt to the input without requesting allfrequency sub-bands provided .…

Modeling the EdNet Dataset with Logistic Regression

We describe our experience with competition from the perspective ofeducational data mining . Many of these challenges are won by neural network models created by artificial intelligence scientists . They have ablack-box character that makes their use and application less clear to learningscientists .…

Measuring the technological pedagogical content knowledge TPACK of in service teachers of computer science who teach algorithms and programming in upper secondary education

This study examines a national sample of 1032 secondaryteachers of computer science and measures their knowledge with respect to technology, pedagogy, content knowledge . Findings indicate that content knowledge and technology knowledge ratingare high . Secondaryteachers are less confident with their pedagogical content knowledge, and theirtechnological content knowledge rating are high .…

Cybernetics and the Future of Work

The disruption caused by the pandemic has called into question industrialnorms and created an opportunity to reimagine the future of work . We discuss how this period of opportunity may be leveraged to bring about a future in which the workforce thrives rather than survives .…

Differentiable SLAM net Learning Particle SLAM for Visual Navigation

Simultaneous localization and mapping (SLAM) remains challenging for a number of downstream applications, such as visual robot navigation, because of rapidturns, featureless walls, and poor camera quality . SLAM-net encodes a particle filter based SLAM algorithm in a differentiablecomputation graph, and learns task-oriented neural network components by backpropagating through the SLAM algorithms .…

Studying the association of online brand importance with museum visitors An application of the semantic brand score

This paper explores the association between brand importance and growth in museum visitors . We analyzed 10 years of online forum discussions and applied the Semantic Brand Score (SBS) to assess the brand importance of five European museums . Results suggest that in order to attract more visitors, museum brand managers should focus on increasing the volume of online posting and the richness of information generated by users around the brand, rather than controlling for the posts’ overall positivity or negativity .…

Choice Set Confounding in Discrete Choice

Standard methods in preference learning involve estimating parameters of choice models from data of selections (choices) made by individuals from a discrete set of alternatives . Ignoring these assignment mechanisms can mislead choice models into making biased estimates of preferences .…

A Data Efficient Approach to Behind the Meter Solar Generation Disaggregation

With the emergence of battery storage and the decline in thesolar photovoltaic (PV) levelized cost of energy (LCOE), the number ofbehind-the-meter solar PV systems is expected to increase steadily . The ability to estimate solar generation from these latent systems is crucial for a range of applications, including distribution system planning and operation, demandresponse, and non-intrusive load monitoring .…

Distributionally Robust Chance Constrained Flexibility Planning for Integrated Energy System

Inflexible combined heat and power plants and uncertain wind power production result in excess power in distribution networks . Power-to-X facilities such aselectrolyser and electric boilers can offer extra flexibility to theintegrated energy system . A case study validates the effectiveness of introducing the electrolyserand electric boiler into the integrated energy system, with respect to the decreased system cost, expanded CHP plant flexibility and reduced inverse powerflow .…

On Decentralization of Bitcoin An Asset Perspective

Since its advent in 2009, Bitcoin, a cryptography-enabled peer-to-peer digital payment system, has been gaining increasing attention from bothacademia and industry . We present in this paper the first systematicinvestigation of the degree of decentralization for Bitcoin based on its entiretransaction history .…

Buying time in software development how estimates become commitments

Despite years of research for improving accuracy, software practitioners still face software estimation difficulties . Researchers’ focus on raising realism inestimates when using it seems not to be enough for the much-expectedimprovements . Instead of focusing on the estimation process’s technicalities, we investigated interaction of the establishment of commitments with customers and software estimation .…

Mining Architecture Tactics and Quality Attributes Knowledge in Stack Overflow

Architecture Tactics (ATs) are architectural building blocks that provide general architectural solutions for addressing Quality Attributes (QAs) issues . Mining and analyzing QA-AT knowledge can help the software architecture community better understand architecture design . Using Stack Overflow (SO) as our source, our main goals are to effectively mine such knowledge; and to have some sense of how developers use ATs with respect to QAconcerns from related discussions .…

DISCO Verification Division of Input Space into COnvex polytopes for neural network verification

The impressive results of modern neural networks partly come from their nonlinear behaviour . We propose amethod to simplify the verification problem by operating a partitionning intomultiple linear subproblems . To evaluate the feasibility of such an approach, we perform an empirical analysis of neural networks to estimate the number of linear regions, and compare them to the bounds currently known .…

Traffic Aware Service Relocation in Cloud Oriented Elastic Optical Networks

In this paper, we study problem of efficient service relocation (i.e.,changing assigned data center for a selected client node) in elastic opticalnetworks . We propose dedicated flow allocation algorithm that can be supported by the service relocation process . We also introduce 21 different relocationpolicies, which use three types of data for decision making – networktopological characteristics, rejection history and traffic prediction .…

StRETcH a Soft to Resistive Elastic Tactile Hand

Soft optical tactile sensors enable robots to manipulate deformable objects . StRETcH, a Soft to Resistive Elastic Tactile Hand, is easily manufactured and integrated with a robotic arm . An elasticmembrane is suspended between two robotic fingers, and a depth sensor captures deformations of the elastic membrane .…

Compacting Squares

Edge-connected configurations of squares are a common model for modularrobots in two dimensions . Dumitrescu and Pach proved that it is always possible to reconfigure oneedge-connected configuration of $n$ squares into any other using at most $O(n^2)$ sliding moves . However, significantly fewer moves may be sufficient .…

On exploration requirements for learning safety constraints

Enforcing safety for dynamical systems is challenging, since it requiresconstraint satisfaction along trajectory predictions . Equivalent controlconstraints can be computed in the form of sets that enforce positiveinvariance . However, these constraints are cumbersome to compute from models, and it is not yet well established how to infer constraints from data .…

Controlling an Inverted Pendulum with Policy Gradient Methods A Tutorial

This paper provides the details of implementing two important policy gradient algorithms to solve the inverted pendulum problem . These are namely the DeepDeterministic Policy Gradient (DDPG) and the Proximal Policy Optimization (PPO) The problem is solved by using an actor-critic model where anactor-network is used to learn the policy function and a critic network is toevaluate the actor-network by learning to estimate the Q function .…

Topology Optimization for Large Scale Additive Manufacturing Generating designs tailored to the deposition nozzle size

Additive Manufacturing (AM) processes intended for large scale components deposit large volumes of material to shorten process duration . This reduces the resolution of the AM process, which is typically defined by the size of the deposition nozzle . This article proposes and compares two methods, which are based on existing TO techniques that enable control of minimum and maximummember size, and of minimum cavity size .…

FOGA Flag Optimization with Genetic Algorithm

Flag Optimization with GeneticAlgorithm (FOGA) is an autotuning solution for GCC flag optimization . FOGA hastwo main advantages: the hyperparameter tuning of the genetic algorithm (GA), the second one is themaximum iteration parameter to stop when no further improvement occurs .…