arXiv subjects

Di Feng

Publications and source records attributed to Di Feng.

43 records · Page 3Linked to original sources

arXiv2019

Deep Active Learning for Efficient Training of a LiDAR 3D Object Detector

Training a deep object detector for autonomous driving requires a huge amount of labeled data. While recording data via on-board sensors such as camera or LiDAR is relatively easy, annotating data is very tedious and time-consuming, especially when dealing with 3D LiDAR points or radar data. Active learning has the potential to minimize human annotation efforts while maximizing the object detector's performance. In this work, we propose an active learning method to train a LiDAR 3D object detector with the least amount of labeled training data necessary. The detector leverages 2D region proposals generated from the RGB images to reduce the search space of objects and speed up the learning process. Experiments show that our proposed method works under different uncertainty estimations and query functions, and can save up to 60% of the labeling efforts while reaching the same network performance.

cs.RO↗

arXiv2019

A Utility-Driven Multi-Queue Admission Control Solution for Network Slicing

The combination of recent emerging technologies such as network function virtualization (NFV) and network programmability (SDN) gave birth to the Network Slicing revolution. 5G networks consist of multi-tenant infrastructures capable of offering leased network "slices" to new customers (e.g., vertical industries) enabling a new telecom business model: Slice-as-aService (SlaaS). In this paper, we aim i ) to study the slicing admission control problem by means of a multi-queuing system for heterogeneous tenant requests, ii ) to derive its statistical behavior model, and iii ) to provide a utility-based admission control optimization. Our results analyze the capability of the proposed SlaaS system to be approximately Markovian and evaluate its performance as compared to legacy solutions.

cs.NI↗

arXiv2018

Towards Safe Autonomous Driving: Capture Uncertainty in the Deep Neural Network For Lidar 3D Vehicle Detection

To assure that an autonomous car is driving safely on public roads, its object detection module should not only work correctly, but show its prediction confidence as well. Previous object detectors driven by deep learning do not explicitly model uncertainties in the neural network. We tackle with this problem by presenting practical methods to capture uncertainties in a 3D vehicle detector for Lidar point clouds. The proposed probabilistic detector represents reliable epistemic uncertainty and aleatoric uncertainty in classification and localization tasks. Experimental results show that the epistemic uncertainty is related to the detection accuracy, whereas the aleatoric uncertainty is influenced by vehicle distance and occlusion. The results also show that we can improve the detection performance by 1%-5% by modeling the aleatoric uncertainty.

cs.RO↗

arXiv2018

A Markov Model of Slice Admission Control

The emerging feature of network slicing in future Fifth Generation (5G) networks calls for efficient slice management. Recent studies have been focusing on the mechanism of slice admission control, which functions in a manner of state machine. This paper proposes a general state model for synchronous slice admission control, and proves it to be Markovian under a set of weak constraints. An analytical approximation of the state transition matrix to reduce computational complexity in practical applications is also proposed and evaluated.

cs.NI↗

arXiv2018

Leveraging Robotic Prior Tactile Exploratory Action Experiences For Learning New Objects's Physical Properties

Reusing the tactile knowledge of some previously-explored objects helps us humans to easily recognize the tactual properties of new objects. In this master thesis, we enable arobotic arm equipped with multi-modal artificial skin, like humans, to actively transfer the prior tactile exploratory action experiences when it learns the detailed physical properties of new objects. These prior tactile experiences are built when the robot applies the pressing, sliding and static contact movements on objects with different action parameters and perceives the tactile feedbacks from multiple sensory modalities. Our method was systematically evaluated by several experiments. Results show that the robot could consistently improve the discrimination accuracy by over 10% when it exploited the prior tactile knowledge compared with using no transfer method, and 25% when it used only one training sample. The results also show that the proposed method was robust against transferring irrelevant prior tactile knowledge.

cs.RO↗

arXiv2017

A Profit-Maximizing Strategy of Network Resource Management for 5G Tenant Slices

Supported by the emerging technologies of Network Function Virtualization (NFV) and network slicing, 5G networks allow tenants to rent resources from mobile network operators (MNOs) in order to provide services without possessing an own network infrastructure. The MNOs are therefore facing the problem of deciding if to accept or decline the resource renting requests they receive. This paper builds a stochastic model that describes the MNO's revenue and opportunity cost of accepting a contract, and therewith proposes a strategy that is analytically derived to maximize the expected profit at every decision.

cs.NI↗

arXiv2014

Plasmonic-photonic crystal coupled nanolaser

We propose and demonstrate a hybrid photonic-plasmonic nanolaser that combines the light harvesting features of a dielectric photonic crystal cavity with the extraordinary confining properties of an optical nano-antenna. In that purpose, we developed a novel fabrication method based on multi-step electron-beam lithography. We show that it enables the robust and reproducible production of hybrid structures, using fully top down approach to accurately position the antenna. Coherent coupling of the photonic and plasmonic modes is highlighted and opens up a broad range of new hybrid nanophotonic devices.

physics.optics↗