This paper addresses theproblem of cascading errors by focusing on the coupling between the trackingand prediction modules . Rather than relying on a single set of tracking results forprediction, our framework simultaneously reasons about multiple sets oftracking results . We show that this framework improves overall prediction performance over the standard single-hypothesistracking-prediction pipeline by up to 34.2% on the nuScenes dataset, with evenmore significant improvements (up to ~70%) when restricting the evaluation to challenging scenarios involving identity switches and fragments — all with anacceptable computation overhead .…
Cutting Voxel Projector a New Approach to Construct 3D Cone Beam CT Operator
In this paper, we introduce a new class of projectors for 3D cone beamtomographic reconstruction . We construct a near-exact projector and backprojector that can beused especially for algebraic reconstruction techniques . We show that the cutting voxel projector achieves, especially for large cone beam angles, noticeably higher accuracy than the TT projector .…
Hermite multiwavelets for manifold valued data
In this paper we present a construction of interpolatory Hermitemultiwavelets for functions that take values in nonlinear geometries such as Riemannian manifolds or Lie groups . We rely on the strong connection between wavelets and subdivision schemes to define a prediction-correction approach .…
Fine Grained Complexity Theory Conditional Lower Bounds for Computational Geometry
Fine-grained complexity theory is the area of theoretical computer sciencethat proves conditional lower bounds based on the Strong Exponential TimeHypothesis and similar conjectures . This area has been thriving in the last decade, leading to conditionally best-possible algorithms for a wide variety of problems on graphs, strings, numbers etc.…
DNA Codes over the Ring mathbb Z _4 w mathbb Z _4
We extend the Gau map andGau distance, defined in DKBG, over all the $16$ rings$\mathcal{R}__\theta . An isometry between the codes over the rings and the analogous DNA codes is established in general . We propose family of DNA codes that satisfies reverse and reverse-complement constraints using the Reed-Muller type codes .…
Capacity Region Bounds for the K user Dispersive Nonlinear Optical WDM Channel with Peak Power Constraints
It is known that fiber nonlinearities induce crosstalk in a wavelengthdivision multiplexed (WDM) system . We characterize, for the first time, an outer bound on the capacity region of simultaneously achievablerate pairs, assuming a general K-user perturbative channel model usinggenie-aided techniques .…
Minimum ell_ 1 norm interpolators Precise asymptotics and multiple descent
An evolving line of machine learning works observe empirical evidence that interpolating estimators may not necessarily be harmful . This paper pursues theoretical understandingfor an important type of interpolator: the minimum $ell_1$-norminterpolator . We observe a curious multi-descent phenomenon; that is, the generalization risk of the minimum$\ell__1 $-norm interpolator undergoes multiple (and possibly more than two)phases of descent and ascent as one increases the model capacity .…
Dynamic Tolling for Inducing Socially Optimal Traffic Loads
How to design tolls that induce socially optimal traffic loads? We propose atwo-timescale discrete-time stochastic dynamics that adaptively adjusts thetoll prices on a parallel link network . We show that the loads and the tolls concentrate in a neighborhood of the fixed point, which correspondsto the socially optimal load and toll price .…
Wideband and Entropy Aware Deep Soft Bit Quantization
Deep learning has been recently applied to physical layer processing indigital communication systems . We introduce a novel deep learning solution for soft bitquantization across wideband channels . Our method is trained end-to-end withquantization- and entropy-aware augmentations to the loss function .…
Turing Tumble is Turing Complete
It is shown that the toy Turing Tumble, suitably extended with an infinitelylong game board and unlimited supply of pieces, is Turing-Complete . This isachieved via direct simulation of a Turing machine . Unlike previouslyinformally presented constructions, we do not encode the finite controlinfinitely many times, we need only one trigger/ball-hopper pair, and we proveour construction correct .…
Fair and Efficient Allocations of Chores under Bivalued Preferences
We study the problem of fair and efficient allocation of a set of indivisible chores to agents with additive cost functions . We consider the popular fairnessnotion of envy-freeness up to one good (EF1) with the efficiency notion ofPareto-optimality (PO) While it is known that an EF1+PO allocation exists andcan be computed in pseudo-polynomial time in the case of goods, the same problem is open for chores .…
Hybrid Layers Neural Network Architectures for Modeling the Self Interference in Full Duplex Systems
Full-duplex (FD) systems have been introduced to provide high data rates forbeyond fifth-generation wireless networks . This article proposes two novel hybrid-layersneural network (NN) architectures to cancel the SI with low complexity . The key idea behindusing hybrid layers is to build an NN model, which makes use of the different layers employed in its architecture .…
Context aware Reranking with Utility Maximization for Recommendation
Reranking rearranges items in the initial rankinglists from the previous ranking stage to better meet users’ demands . An ideal reranking algorithm should consider the counterfactual context — the position and the alignment of the items in reranked lists . CRUM significantly outperforms the state-of-the-art models in terms of both relevance-based metrics and utility-based metric metrics.…
Low Precision Quantization for Efficient Nearest Neighbor Search
K-Nearest Neighbor search over real-valued vector spaces (KNN) is animportant algorithmic task for information retrieval and recommendationsystems . We present a method for using reduced precision to represent vectorsthrough quantized integer values, enabling both a reduction in the memoryoverhead of indexing these vectors and faster distance computations at querytime .…
EmbRace Accelerating Sparse Communication for Distributed Training of NLP Neural Networks
EmbRace introduces Sparsity-aware Hybrid Communication, which combines AlltoAlland AllReduce to optimize communication overhead for sparse and dense datain NLP models . EmbRace achieves up to 30.66X speedupon 16 GPUs clusters among four popular distributed training baselines . Experimental results show that EmbRace can be achieved with four representative NLP model models on two high-performance GPU clusters .…
Ctrl Shift How Privacy Sentiment Changed from 2019 to 2021
People’s privacy sentiments drive changes in legislation and may influencetheir willingness to use a variety of technologies . After the onset of COVID-19, we observe significant changes in Americans’ privacy sentimentstoward government- and health-related data uses . We observe additionalchanges in the context of other national events such as the U.S.…
A cautionary tale on fitting decision trees to data from additive models generalization lower bounds
Decision trees are important both as interpretable models amenable to high-stakes decision-making, and as building blocks of ensemble methods such asrandom forests and gradient boosting . Their statistical properties, however,are not well understood . We prove a sharpsquared error generalization lower bound for a large class of decision tree algorithms fitted to sparse additive models with $C^1$ component functions .…
Comparing Deep Neural Nets with UMAP Tour
A tool, UMAP Tour, is built to visually inspect and compare behavior of real-world neural network models . The method used in the visualization alsoimplies a new similarity measure between neural network layers . Using the tool and the similarity measure, we find concepts learned instate-of-the-art models and dissimilarities between them, such as GoogLeNet andResNet .…
Understanding Players Interaction Patterns with Mobile Game App UI via Visualizations
Understanding how players interact with the mobile game app on smartphonedevices is important for game experts to develop and refine their app products . Visualizing therecorded logs of users’ UI operations is a promising way for quantitativelyunderstanding the interaction patterns .…
Uncertainty aware Topic Modeling Visualization
Topic modeling is a state-of-the-art technique for analyzing text corpora . It uses a statistical model, most commonly Latent Dirichlet Allocation (LDA) to discover abstract topics that occur in the document collection . The LDA-based topic modeling procedure is based on a randomly selected initialconfiguration .…
Affine Hermitian Grassmann Codes
The Grassmannian is an important object in Algebraic Geometry . We introduce a new class of linear codes called Affine Hermitian GrassmanCodes . These codes are the linear codes resulting from an affine part of the projective embedding of the Grassmann codes .…
Empirical Policy Optimization for n Player Markov Games
In single-agent Markov decision processes, an agent can optimize its policy based on the interaction with environment . In multi-player Markov games, the interaction is non-stationary due to the behaviors of other players . The core is to evolve one’s policy according to not just its current in-game performance, but an aggregation of its performance over history .…
Fairness Concepts for Indivisible Items with Externalities
We study a fair allocation problem of indivisible items under additiveexternalities in which each agent also receives values from items that are assigned to other agents . We propose a new fairness concept called general fair share (GFS) We undertake a detailed study and present algorithms for finding fair allocations .…
Bayesian Persuasion in Sequential Trials
We consider a Bayesian persuasion or information design problem where thesender tries to persuade the receiver to take a particular action via a sequence of signals . This we model by considering multi-phase trials where the FDA determines some of the experiments .…
Online Motion Planning with Soft Timed Temporal Logic in Dynamic and Unknown Environment
Motion planning of an autonomous system with high-level specifications has wide applications . Research of formal languages involving timedtemporal logic is still under investigation . Many existing resultsrely on a key assumption that user-specified tasks are feasible in the given environment .…
Learning to Learn a Cold start Sequential Recommender
The cold-start recommendation is an urgent problem in contemporary online applications . Many data-driven algorithms, suchas the widely used matrix factorization, underperform because of datasparseness . This work adopts the idea of meta-learning to solve the user’scold start recommendation problem .…
Small Data and Process in Data Visualization The Radical Translations Case Study
This paper uses the collaborative project Radical Translations as case study to examine some of the theoretical perspectives informing the adoption andcritique of data visualization in the digital humanities . It showcases how data visualization is used within a King’s DigitalLab project lifecycle to facilitate collaborative data exploration within the project interdisciplinary team .…
Ranking Facts for Explaining Answers to Elementary Science Questions
Explanations are created from a human-annotated set ofnearly 5,000 candidate facts in the WorldTree corpus . Our aim is to obtainbetter matches for valid facts of an explanation for the correct answer of a question over the available fact candidates .…
Deep Learning Based Power Control for Uplink Cell Free Massive MIMO Systems
A framework for deep learning-based power control methods for max-min, max-product and max-sum-rate optimization in uplinkcell-free massive multiple-input multiple- input multiple-output (CF mMIMO) systems is proposed . Instead of using supervised learning, the proposed method relies onunsupervised learning, in which optimal power allocations are not required to be known, and thus has low training complexity .…
Newsalyze Effective Communication of Person Targeting Biases in News Articles
Media bias and its extreme form, fake news, can decisively affect public opinion, authors say . Slanted news coverage may influence societal decisions, e.g., in democratic elections . Study suggests that our content-driven identification method detects groups of similarly slanted news articles due to substantial biases present in individual news articles.…
Semantics of Conjectures
This paper aims to expand and detail the notion of formal semantics of Conjectures by applying a theoretic-model approach . After a short introduction to the concepts and basics, we will start from the concept ofSimple Interpretation of RDF, applying and extending the semantic rules and conditions to fully cover the concepts .…
Location Information Assisted Beamforming Design for Reconfigurable Intelligent Surface Aided Communication Systems
In reconfigurable intelligent surface (RIS) aided millimeter-wave (mmWave)communication systems, in order to overcome the limitation of the conventional channel state information (CSI) acquisition techniques, this paper proposes alocation information assisted beamforming design without the requirement of the channel training process .…
Ride Sharing Data Privacy An Analysis of the State of Practice
Digital services like ride sharing rely heavily on personal data . Services include a varying set of personal data and offer limited privacy-related features . Privacy concerns are a decisive factor for individuals to (not) use these services . The results show that services include a different set of data and are often limited on privacy- related features .…
RL4RS A Real World Benchmark for Reinforcement Learning based Recommender System
Reinforcement learning based recommender systems (RL-based RS) aims at learning a good policy from a batch of collected data . However, current RL-basedRS benchmarks commonly have a large reality gap, because they involve artificial RL datasets or semi-simulated RS datasets .…
DNA Codes over the Ring mathbb Z _4 w mathbb Z _4
We extend the Gau map andGau distance, defined in DKBG, over all the $16$ rings$\mathcal{R}__\theta . An isometry between the codes over the rings and the analogous DNA codes is established in general . We propose family of DNA codes that satisfies reverse and reverse-complement constraints using the Reed-Muller type codes .…
Ctrl Shift How Privacy Sentiment Changed from 2019 to 2021
People’s privacy sentiments drive changes in legislation and may influencetheir willingness to use a variety of technologies . After the onset of COVID-19, we observe significant changes in Americans’ privacy sentimentstoward government- and health-related data uses . We observe additionalchanges in the context of other national events such as the U.S.…
Model Order Estimation for A Sum of Complex Exponentials
In this paper, we present a new method for estimating the number of terms in a sum of exponentially damped sinusoids embedded in noise . We propose to combine the shift-invariance property of the Hankel matrix with a constraint over its singular values topenalize small order estimations .…
Joint Spatial Division and Coaxial Multiplexing for Downlink Multi User OAM Wireless Backhaul
Orbital angular momentum (OAM) at radio frequency (RF) provides a novel approach of multiplexing a set of orthogonal modes on the same frequency channel to achieve high spectral efficiencies (SEs) At last, the proposed methods are extended to the downlink MU-OAM-MIMO wireless backhaul system equipped with uniform concentric circular arrays (UCCAs) for which much higher spectral efficiency (SE) and energy efficiency (EE) can be achieved .…
Reconfigurable Intelligent Surface Enhanced OFDM Communications via Delay Adjustable Metasurface
Reconfigurable intelligent surface (RIS) is a promising technology forestablishing spectral- and energy-efficient wireless networks . In this paper, we study RIS-enhanced orthogonal frequency division multiplexing (OFDM)communications . We introduce the delay adjustablemetasurface (DAM) relying on varactor diodes . In contrast to existingreflecting elements, each one in DAM is capable of storing and retrieving theimpinging electromagnetic waves upon dynamically controlling itselectromagnetically induced transparency (EIT) properties .…
Projected Model Counting Beyond Independent Support
The past decade has witnessed a surge of interest in practical techniques forprojected model counting . Despite significant advancements, performance scaling remains the Achilles’ heel of this field . Inapplications such as verification of binarized neural networks, quantification of information flow, reliability of power grids etc.,…
Learning Realtime One Counter Automata
Ouralgorithm uses membership and equivalence queries as in Angluin’s L* algorithm . In a partialequivalence query, we ask the teacher whether the language of a givenfinite-state automaton coincides with a counter-bounded subset of the targetlanguage . We evaluate an implementation of our algorithm on a number of randombenchmarks .…
Sequential Modeling with Multiple Attributes for Watchlist Recommendation in E Commerce
In e-commerce, the watchlist enables users to track items over time and hasemerged as a primary feature, playing an important role in users’ shopping journey . Watchlist items typically have multiple attributes whose values maychange over time (e.g., price, quantity) Since many users accumulate dozens of items on their watchlist, and since shopping intents change over time,recommending the top watchlist items in a given context can be valuable .…
Don t Judge Me by My Face An Indirect Adversarial Approach to Remove Sensitive Information From Multimodal Neural Representation in Asynchronous Job Video Interviews
Adversarial methods have been proven to effectively remove sensitive information from the latent representation of neural networks without the need to collect any sensitive variable . This is the first application of adversarial techniques for obtaining a multimodal fair representation in the context of video job interviews .…
DroneStick Flying Joystick as a Novel Type of Interface
DroneStick is a novel hands-free method for smooth interaction between a human and a robotic system via one of its agents . A flying joystick(DroneStick) is composed of a flying drone and coiled wire with a vibration motor . A potential application can enhance an automated `lastmile’ delivery when a recipient needs to guide a delivery drone/robot gently to a spot where a parcel has to be dropped .…
Structured vector fitting framework for mechanical systems
In this paper, we develop a structure-preserving formulation of the data-driven vector fitting algorithm for the case of modally damped mechanical systems . We propose two possible structuredextensions of the barycentric formula of system transfer functions . Integrating these new forms within the classical vector fitting algorithms leads to theformulation of two new algorithms .…
How to Effectively Identify and Communicate Person Targeting Media Bias in Daily News Consumption
Slanted news coverage has long been studied in the socialsciences, resulting in comprehensive models to describe it and effective yetcostly methods to analyze it . We present anin-progress system for news recommendation that is the first to automate themanual procedure of content analysis to reveal person-targeting biases in news articles reporting on policy issues .…
Measuring Cognitive Status from Speech in a Smart Home Environment
By 2050, one in six people in the world will be over age 65 (up from one in 11 in 2019) Smart devices and smart home technology have potential to transform how people age, their ability to live independently in later years .…
A Primer on the Statistical Relation between Wireless Ultra Reliability and Location Estimation
This letter statistically characterizes the impact of location estimationuncertainty in the wireless communication reliability . The reliability – characterized by how likely the outage probabilityis to be above a target threshold – can be sensitive to location errors . We highlight the difficulty of choosing a rate that both meets targetreliability and accounts for the location uncertainty, and that the most directsolutions suffer from being too conservative .…
The search of Type I codes
A self-dual binary linear code is called Type I code if it has singly-evencodewords, i.e.~it has codewords with weight divisible by $2.$ The purpose of this paper is to investigate interesting properties of Type I codes . Further, we build up a computer-based code-searching program .…
A Formalisation of Abstract Argumentation in Higher Order Logic
We present an approach for representing abstract argumentation frameworks based on an encoding into classical higher-order logic . This provides a uniform framework for computer-assisted assessment of abstract argumentations . This enables the formalanalysis and verification of meta-theoretical properties as well as theflexible generation of extensions and labellings with respect to well-known semantics .…