References of "Ottersten, Björn 50002797"
     in
Bookmark and Share    
Full Text
Peer Reviewed
See detailDeformation Based 3D Facial Expression Representation
Demisse, Girum UL; Aouada, Djamila UL; Ottersten, Björn UL

in ACM Transactions on Multimedia Computing, Communications, & Applications (2018)

We propose a deformation based representation for analyzing expressions from 3D faces. A point cloud of a 3D face is decomposed into an ordered deformable set of curves that start from a fixed point ... [more ▼]

We propose a deformation based representation for analyzing expressions from 3D faces. A point cloud of a 3D face is decomposed into an ordered deformable set of curves that start from a fixed point. Subsequently, a mapping function is defined to identify the set of curves with an element of a high dimensional matrix Lie group, specifically the direct product of SE(3). Representing 3D faces as an element of a high dimensional Lie group has two main advantages. First, using the group structure, facial expressions can be decoupled from a neutral face. Second, an underlying non-linear facial expression manifold can be captured with the Lie group and mapped to a linear space, Lie algebra of the group. This opens up the possibility of classifying facial expressions with linear models without compromising the underlying manifold. Alternatively, linear combinations of linearised facial expressions can be mapped back from the Lie algebra to the Lie group. The approach is tested on the BU-3DFE and the Bosphorus datasets. The results show that the proposed approach performed comparably, on the BU-3DFE dataset, without using features or extensive landmark points. [less ▲]

Detailed reference viewed: 263 (27 UL)
Full Text
Peer Reviewed
See detailOFDM-based automotive joint radar-communication system
Dokhanchi, Sayed Hossein UL; Shankar, Bhavani UL; Stifter, Thomas et al

in 2018 IEEE Radar Conference (RadarConf18) (2018)

We propose a novel automotive joint radar-communication (JRC) system, where the system first transmits OFDM sub-carriers for radar processing followed by sub-carriers enabling radar and communication ... [more ▼]

We propose a novel automotive joint radar-communication (JRC) system, where the system first transmits OFDM sub-carriers for radar processing followed by sub-carriers enabling radar and communication functionalities. The receiver processing includes iterative estimation of parameters to alleviate the shortage of samples to estimate range. The receiver first estimates the target parameters from the sub-carriers dedicated to radar; these parameters then determine the channel for the communication link. The communication data is then extracted, thereby enabling the use of all the carriers for improving the range estimation. It is shown that the range estimation improves significantly after efficient use of all the sub-carriers. Furthermore, for radar parameter estimation, we propose an effective iterative method based on alternating least square (ALS) to recover the angle of arrival (AoA), Doppler and Range. Numerical results demonstrate the feasibility of our proposed system. [less ▲]

Detailed reference viewed: 175 (12 UL)
Full Text
Peer Reviewed
See detail3DBodyTex: Textured 3D Body Dataset
Saint, Alexandre Fabian A UL; Ahmed, Eman UL; Shabayek, Abd El Rahman UL et al

in 2018 Sixth International Conference on 3D Vision (3DV 2018) (2018)

In this paper, a dataset, named 3DBodyTex, of static 3D body scans with high-quality texture information is presented along with a fully automatic method for body model fitting to a 3D scan. 3D shape ... [more ▼]

In this paper, a dataset, named 3DBodyTex, of static 3D body scans with high-quality texture information is presented along with a fully automatic method for body model fitting to a 3D scan. 3D shape modelling is a fundamental area of computer vision that has a wide range of applications in the industry. It is becoming even more important as 3D sensing technologies are entering consumer devices such as smartphones. As the main output of these sensors is the 3D shape, many methods rely on this information alone. The 3D shape information is, however, very high dimensional and leads to models that must handle many degrees of freedom from limited information. Coupling texture and 3D shape alleviates this burden, as the texture of 3D objects is complementary to their shape. Unfortunately, high-quality texture content is lacking from commonly available datasets, and in particular in datasets of 3D body scans. The proposed 3DBodyTex dataset aims to fill this gap with hundreds of high-quality 3D body scans with high-resolution texture. Moreover, a novel fully automatic pipeline to fit a body model to a 3D scan is proposed. It includes a robust 3D landmark estimator that takes advantage of the high-resolution texture of 3DBodyTex. The pipeline is applied to the scans, and the results are reported and discussed, showcasing the diversity of the features in the dataset. [less ▲]

Detailed reference viewed: 1092 (75 UL)
Full Text
Peer Reviewed
See detailHybrid analog-digital transceiver designs for mmwave amplify-and-forward relaying systems
Tsinos, Christos UL; Chatzinotas, Symeon UL; Ottersten, Björn UL

in 2018 41st International Conference on Telecommunications and Signal Processing (TSP) (2018)

Hybrid Analog/Digital (A/D) pre-/post-coding solutions aim at the reduction of the hardware complexity and power consumption of a system employed with a large-scale antenna array functioning in the ... [more ▼]

Hybrid Analog/Digital (A/D) pre-/post-coding solutions aim at the reduction of the hardware complexity and power consumption of a system employed with a large-scale antenna array functioning in the millimeter (mmWave) band. This is achieved by enabling the transceiver design with fewer Radio Frequency (RF) chains than antennas. In this work, hybrid A/D transceiver designs are developed for a relay assisted mmWave system that aims at data transmission from a source to a destination node via the well-known amplify-and-forward cooperative protocol. To that end, hybrid solutions are proposed for the joint design of the source and relay nodes precoders and for the destination ones, as well. Contrariwise to existing literature approaches that develop codebook-based solutions and exhibit limitations on the performance, the proposed approaches are codebook-free and present significantly improved spectral efficiency, as it is verified via the presented simulations. [less ▲]

Detailed reference viewed: 102 (2 UL)
Full Text
Peer Reviewed
See detailBinary Sequences set with small ISL for MIMO radar systems
Alaee-Kerahroodi, Mohammad; Modarres-Hashemi, Mahmoud; Naghsh, Mohammad Mahdi Naghsh et al

in 2018 26th European Signal Processing Conference (EUSIPCO) (2018)

In this paper, we aim at designing a set of binary sequences with good aperiodic auto- and crosscorrelation properties for Multiple-Input-Multiple-Output (MIMO) radar systems. We show such a set of ... [more ▼]

In this paper, we aim at designing a set of binary sequences with good aperiodic auto- and crosscorrelation properties for Multiple-Input-Multiple-Output (MIMO) radar systems. We show such a set of sequences can be obtained by minimizing the Integrated Side Lobe (ISL) with the binary requirement imposed as a design constraint. By using the block coordinate descent (BCD) framework, we propose an efficient monotonic algorithm based on Fast Fourier Transform (FFT), to minimize the objective function which is non-convex and NP-hard in general. Simulation results illustrate that the ISL of designed binary set of sequences is the neighborhood of the Welch bound, Indicating its superior performance. [less ▲]

Detailed reference viewed: 127 (4 UL)
Full Text
Peer Reviewed
See detailA Revisit of Action Detection using Improved Trajectories
Papadopoulos, Konstantinos UL; Antunes, Michel; Aouada, Djamila UL et al

in IEEE International Conference on Acoustics, Speech and Signal Processing, Calgary, Alberta, Canada, 15–20 April 2018 (2018)

In this paper, we revisit trajectory-based action detection in a potent and non-uniform way. Improved trajectories have been proven to be an effective model for motion description in action recognition ... [more ▼]

In this paper, we revisit trajectory-based action detection in a potent and non-uniform way. Improved trajectories have been proven to be an effective model for motion description in action recognition. In temporal action localization, however, this approach is not efficiently exploited. Trajectory features extracted from uniform video segments result in significant performance degradation due to two reasons: (a) during uniform segmentation, a significant amount of noise is often added to the main action and (b) partial actions can have negative impact in classifier's performance. Since uniform video segmentation seems to be insufficient for this task, we propose a two-step supervised non-uniform segmentation, performed in an online manner. Action proposals are generated using either 2D or 3D data, therefore action classification can be directly performed on them using the standard improved trajectories approach. We experimentally compare our method with other approaches and we show improved performance on a challenging online action detection dataset. [less ▲]

Detailed reference viewed: 236 (27 UL)
Full Text
Peer Reviewed
See detailUser Selection for symbol-level multigroup multicasting precoding in the downlink of MISO channels
Alodeh, Maha UL; Chatzinotas, Symeon UL; Ottersten, Björn UL

in 2018 IEEE International Conference on Communications (ICC) (2018)

We consider the problem of user selection for symbol-level multigroup multicasting in the downlink of multiuser MISO systems. Symbol-level precoding is a new paradigm for multiuser multiple-antenna ... [more ▼]

We consider the problem of user selection for symbol-level multigroup multicasting in the downlink of multiuser MISO systems. Symbol-level precoding is a new paradigm for multiuser multiple-antenna downlink systems which aims at creating constructive interference among the simultaneous data streams. This can be enabled by designing the precoded signal of the multiantenna transmitter on a symbol level, taking into account both channel state information and data symbols. This work proposes a user selection algorithm to facilitate serving multiple groups of users, by transmitting a stream of common symbols to each group on symbol-by-symbol basis if we have large number of users. We provide numerical results to validate the proposed algorithm. [less ▲]

Detailed reference viewed: 109 (12 UL)
Full Text
Peer Reviewed
See detailConstrained Bayesian Active Learning of Interference Channels in Cognitive Radio Networks
Tsakmalis, Anestis UL; Chatzinotas, Symeon UL; Ottersten, Björn UL

in IEEE Journal of Selected Topics in Signal Processing (2017)

In this paper, a sequential probing method for interference constraint learning is proposed to allow a centralized Cognitive Radio Network (CRN) accessing the frequency band of a Primary User (PU) in an ... [more ▼]

In this paper, a sequential probing method for interference constraint learning is proposed to allow a centralized Cognitive Radio Network (CRN) accessing the frequency band of a Primary User (PU) in an underlay cognitive scenario with a designed PU protection specification. The main idea is that the CRN probes the PU and subsequently eavesdrops the reverse PU link to acquire the binary ACK/NACK packet. This feedback indicates whether the probing-induced interference is harmful or not and can be used to learn the PU interference constraint. The cognitive part of this sequential probing process is the selection of the power levels of the Secondary Users (SUs) which aims to learn the PU interference constraint with a minimum number of probing attempts while setting a limit on the number of harmful probing-induced interference events or equivalently of NACK packet observations over a time window. This constrained design problem is studied within the Active Learning (AL) framework and an optimal solution is derived and implemented with a sophisticated, accurate and fast Bayesian Learning method, the Expectation Propagation (EP). The performance of this solution is also demonstrated through numerical simulations and compared with modified versions of AL techniques we developed in earlier work. [less ▲]

Detailed reference viewed: 197 (25 UL)
Full Text
Peer Reviewed
See detailSymbol-level Precoding for the Non-linear Multiuser MISO Downlink Channel
Spano, Danilo UL; Alodeh, Maha; Chatzinotas, Symeon UL et al

in IEEE Transactions on Signal Processing (2017)

This paper investigates the problem of the interference among multiple simultaneous transmissions in the downlink channel of a multi-antenna wireless system. A symbol-level precoding scheme is considered ... [more ▼]

This paper investigates the problem of the interference among multiple simultaneous transmissions in the downlink channel of a multi-antenna wireless system. A symbol-level precoding scheme is considered, in order to exploit the multi-user interference and transform it into useful power at the receiver side, through a joint utilization of the data information and the channel state information. In this context, this paper presents novel strategies which exploit the potential of symbol-level precoding to control the per-antenna instantaneous transmit power. In particular, the power peaks amongst the transmitting antennas and the instantaneous power imbalances across the different transmitted streams are minimized. These objectives are particularly relevant with respect to the non-linear amplitude and phase distortions induced by the per-antenna amplifiers, which are important sources of performance degradation in practical systems. More specifically, this work proposes two different symbol-level precoding approaches. A first approach performs a weighted per-antenna power minimization, under Quality-of-Service constraints and under a lower bound constraint on the per-antenna transmit power. A second strategy performs a minimization of the spatial peak-to-average power ratio, evaluated amongst the transmitting antennas. Numerical results are presented in a comparative fashion to show the effectiveness of the proposed techniques, which outperform the state of the art symbol-level precoding schemes in terms of spatial peak-to-average power ratio, spatial dynamic range, and symbol-error-rate over non-linear channels. [less ▲]

Detailed reference viewed: 194 (15 UL)
Full Text
Peer Reviewed
See detailA Framework for Optimizing Multi-cell NOMA: Delivering Demand with Less Resource
You, Lei; Lei, Lei UL; Yuan, Di et al

in 2017 IEEE Global Communications Conference (GLOBECOM) (2017, December)

Non-orthogonal multiple access (NOMA) allows multiple users to simultaneously access the same time-frequency resource by using superposition coding and successive interference cancellation (SIC). Thus far ... [more ▼]

Non-orthogonal multiple access (NOMA) allows multiple users to simultaneously access the same time-frequency resource by using superposition coding and successive interference cancellation (SIC). Thus far, most papers on NOMA have focused on performance gain for one or sometimes two base stations. In this paper, we study multi-cell NOMA and provide a general framework for user clustering and power allocation, taking into account inter-cell interference, for optimizing resource allocation of NOMA in multi-cell networks of arbitrary topology. We provide a series of theoretical analysis, to algorithmically enable optimization approaches. The resulting algorithmic notion is very general. Namely, we prove that for any performance metric that monotonically increases in the cells’ resource consumption, we have convergence guarantee for global optimum. We apply the framework with its algorithmic concept to a multi-cell scenario to demonstrate the gain of NOMA in achieving significantly higher efficiency. [less ▲]

Detailed reference viewed: 217 (23 UL)
Full Text
Peer Reviewed
See detailCache-Assisted Hybrid Satellite-Terrestrial Backhauling for 5G Cellular Networks
Kalantari, Ashkan; Fittipaldi, Marilena; Chatzinotas, Symeon UL et al

in Proceedings of IEEE Global Communications Conference (2017, December)

Fast growth of Internet content and availability of electronic devices such as smart phones and laptops has created an explosive content demand. As one of the 5G technology enablers, caching is a ... [more ▼]

Fast growth of Internet content and availability of electronic devices such as smart phones and laptops has created an explosive content demand. As one of the 5G technology enablers, caching is a promising technique to off-load the network backhaul and reduce the content delivery delay. Satellite communications provides immense area coverage and high data rate, hence, it can be used for large-scale content placement in the caches. In this work, we propose using hybrid mono/multi-beam satellite-terrestrial backhaul network for off-line edge caching of cellular base stations in order to reduce the traffic of terrestrial network. The off-line caching approach is comprised of content placement and content delivery phases. The content placement phase is performed based on local and global content popularities assuming that the content popularity follows Zipf-like distribution. In addition, we propose an approach to generate local content popularities based on a reference Zipf-like distribution to keep the correlation of content popularity. Simulation results show that the hybrid satellite-terrestrial architecture considerably reduces the content placement time while sustaining the cache hit ratio quite close to the upper-bound compared to the satellite-only method. [less ▲]

Detailed reference viewed: 236 (4 UL)
Full Text
Peer Reviewed
See detailEnergy Minimization for Cache-assisted Content Delivery Networks with Wireless Backhaul
Vu, Thang Xuan UL; Chatzinotas, Symeon UL; Ottersten, Björn UL et al

in IEEE Wireless Communications Letters (2017)

Content caching is an efficient technique to reduce delivery latency and system congestion during peak-traffic time by bringing data closer to end users. In this paper, we investigate energy-efficiency ... [more ▼]

Content caching is an efficient technique to reduce delivery latency and system congestion during peak-traffic time by bringing data closer to end users. In this paper, we investigate energy-efficiency performance of cache-assisted content delivery networks with wireless backhaul by taking into account cache capability when designing the signal transmission. We consider multi-layer caching and the performance in cases when both base station (BS) and users are capable of storing content data in their local cache. Specifically, we analyse energy consumption in both backhaul and access links under two uncoded and coded caching strategies. Then two optimization problems are formulated to minimize total energy cost for the two caching strategies while satisfying some given quality of service constraint. We demonstrate via numerical results that the uncoded caching achieves higher energy efficiency than the coded caching in the small user cache size regime. [less ▲]

Detailed reference viewed: 135 (5 UL)
Full Text
Peer Reviewed
See detailCoverage Extension via Side-Lobe Transmission in Multibeam Satellite System
Gharanjik, Ahmad UL; Kmieciak, Jarek; Shankar, Bhavani UL et al

in 23rd Ka and Broadband Communications Conference (2017, October 16)

In this paper, we study feasibility of coverage extension of a multibeam satellite network by providing low-rate communications to terminals located outside the coverage of main beams. Focusing on the MEO ... [more ▼]

In this paper, we study feasibility of coverage extension of a multibeam satellite network by providing low-rate communications to terminals located outside the coverage of main beams. Focusing on the MEO satellite network, and using realistic link budgets from O3b networks, we investigate the performance of both forward and return-links for terminals stationed in the side lobes of the main beams. Particularly, multi-carrier transmission for forward-link and single carrier transmission for return-link are examined and the resulting coverage and data rate for different setups are evaluated. Simulation results verifies that side-lobe transmission can extend the coverage area and provide considerable data rate, thereby providing a solution for enhancing capacity of existing networks. [less ▲]

Detailed reference viewed: 100 (7 UL)
Full Text
Peer Reviewed
See detailSecrecy Analysis of Random Wireless Networks with Multiple Eavesdroppers
Vuppala, Satyanarayana UL; Chatzinotas, Symeon UL; Ottersten, Björn UL

in Proceedings of IEEE Inter. Symp. on Personal, Indoor and Mobile Radio Communications (PIMRC), Montreal, Canada (2017)

Detailed reference viewed: 163 (4 UL)
Full Text
Peer Reviewed
See detailComputationally Efficient Symbol-Level Precoding Communications Demonstrator
Merlano Duncan, Juan Carlos UL; Krivochiza, Jevgenij UL; Andrenacci, Stefano UL et al

in IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (2017, October)

We present a precoded multi-user communication test-bed to demonstrate forward link interference mitigation techniques in a multi-beam satellite system scenario which will enable a full frequency reuse ... [more ▼]

We present a precoded multi-user communication test-bed to demonstrate forward link interference mitigation techniques in a multi-beam satellite system scenario which will enable a full frequency reuse scheme. The developed test-bed provides an end-to-end precoding demonstration, which includes a transmitter, a multi-beam satellite channel emulator and user receivers. Each of these parts can be reconfigured accordingly to the desired test scenario. Precoded communications allow full frequency reuse in multiple-input multiple-output (MIMO) channel environments, where several coordinated antennas simultaneously transmit to a number of independent receivers. The developed real-time transmission test-bed assist in demonstrating, designing and benchmarking of the new Symbol-Level Precoding (SLP) techniques, where the data information is used, along with the channel state information, in order to exploit the multi-user interference and transform it into useful power at the receiver side. The demonstrated SLP techniques are designed in order to be computationally efficient, and can be generalized to others multi-channel interference scenarios. [less ▲]

Detailed reference viewed: 208 (32 UL)
Full Text
Peer Reviewed
See detailRelay Selection Strategies for SWIPT-Enabled Cooperative Wireless Systems
Gautam, Sumit UL; Lagunas, Eva UL; Sharma, Shree Krishna UL et al

in IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC), Montreal, Canada, Oct. 2017 (2017, October)

Detailed reference viewed: 439 (54 UL)
Full Text
Peer Reviewed
See detailTowards Automatic Human Body Model Fitting to a 3D Scan
Saint, Alexandre Fabian A UL; Shabayek, Abd El Rahman UL; Aouada, Djamila UL et al

in D'APUZZO, Nicola (Ed.) Proceedings of 3DBODY.TECH 2017 - 8th International Conference and Exhibition on 3D Body Scanning and Processing Technologies, Montreal QC, Canada, 11-12 Oct. 2017 (2017, October)

This paper presents a method to automatically recover a realistic and accurate body shape of a person wearing clothing from a 3D scan. Indeed, in many practical situations, people are scanned wearing ... [more ▼]

This paper presents a method to automatically recover a realistic and accurate body shape of a person wearing clothing from a 3D scan. Indeed, in many practical situations, people are scanned wearing clothing. The underlying body shape is thus partially or completely occluded. Yet, it is very desirable to recover the shape of a covered body as it provides non-invasive means of measuring and analysing it. This is particularly convenient for patients in medical applications, customers in a retail shop, as well as in security applications where suspicious objects under clothing are to be detected. To recover the body shape from the 3D scan of a person in any pose, a human body model is usually fitted to the scan. Current methods rely on the manual placement of markers on the body to identify anatomical locations and guide the pose fitting. The markers are either physically placed on the body before scanning or placed in software as a postprocessing step. Some other methods detect key points on the scan using 3D feature descriptors to automate the placement of markers. They usually require a large database of 3D scans. We propose to automatically estimate the body pose of a person from a 3D mesh acquired by standard 3D body scanners, with or without texture. To fit a human model to the scan, we use joint locations as anchors. These are detected from multiple 2D views using a conventional body joint detector working on images. In contrast to existing approaches, the proposed method is fully automatic, and takes advantage of the robustness of state-of-art 2D joint detectors. The proposed approach is validated on scans of people in different poses wearing garments of various thicknesses and on scans of one person in multiple poses with known ground truth wearing close-fitting clothing. [less ▲]

Detailed reference viewed: 314 (36 UL)
Full Text
Peer Reviewed
See detailFacial Expression Recognition via Joint Deep Learning of RGB-Depth Map Latent Representations
Oyedotun, Oyebade UL; Demisse, Girum UL; Shabayek, Abd El Rahman UL et al

in 2017 IEEE International Conference on Computer Vision Workshop (ICCVW) (2017, August 21)

Humans use facial expressions successfully for conveying their emotional states. However, replicating such success in the human-computer interaction domain is an active research problem. In this paper, we ... [more ▼]

Humans use facial expressions successfully for conveying their emotional states. However, replicating such success in the human-computer interaction domain is an active research problem. In this paper, we propose deep convolutional neural network (DCNN) for joint learning of robust facial expression features from fused RGB and depth map latent representations. We posit that learning jointly from both modalities result in a more robust classifier for facial expression recognition (FER) as opposed to learning from either of the modalities independently. Particularly, we construct a learning pipeline that allows us to learn several hierarchical levels of feature representations and then perform the fusion of RGB and depth map latent representations for joint learning of facial expressions. Our experimental results on the BU-3DFE dataset validate the proposed fusion approach, as a model learned from the joint modalities outperforms models learned from either of the modalities. [less ▲]

Detailed reference viewed: 373 (55 UL)
Full Text
Peer Reviewed
See detailRandom Phase Center Motion Technique for Enhanced Angle-Doppler Discrimination Using MIMO Radars
Hammes, Christian UL; Shankar, Bhavani UL; Nijsure, Yogesh UL et al

in European Signal Processing Conference (EUSIPCO) 2017 (2017, August)

A random Phase Center Motion (PCM) technique is presented in this paper, based on Frequency Modulated Continuous Wave (FMCW) radar, in order to suppress the angle- Doppler coupling in Time Division ... [more ▼]

A random Phase Center Motion (PCM) technique is presented in this paper, based on Frequency Modulated Continuous Wave (FMCW) radar, in order to suppress the angle- Doppler coupling in Time Division Multiplex (TDM) Multiple- Input-Multiple-Output (MIMO) radar when employing sparse array structures. The presented approach exploits an apparently moving transmit platform or PCM due to spatio-temporal transmit array modulation. In particular, the work considers a framework utilizing a random PCM trajectory. The statistical characterization of the random PCM trajectory is devised, such that the PCM and the target motion coupling is minimal, while the angular resolution is increased by enabling the virtual MIMO concept. In more details, this paper discusses sidelobe suppression approaches within the angle-Doppler Ambiguity Function (AF) by introducing a phase center probability density function within the array. This allows for enhanced discrimination of multiple targets. Simulation results demonstrate the suppression angle- Doppler coupling by more than 30 dB, even though spatiotemporal transmit array modulation is done across chirps which leads usually to strong angle-Doppler coupling. [less ▲]

Detailed reference viewed: 243 (17 UL)
Full Text
Peer Reviewed
See detailTraining Very Deep Networks via Residual Learning with Stochastic Input Shortcut Connections
Oyedotun, Oyebade UL; Shabayek, Abd El Rahman UL; Aouada, Djamila UL et al

in 24th International Conference on Neural Information Processing, Guangzhou, China, November 14–18, 2017 (2017, July 31)

Many works have posited the benefit of depth in deep networks. However, one of the problems encountered in the training of very deep networks is feature reuse; that is, features are ’diluted’ as they are ... [more ▼]

Many works have posited the benefit of depth in deep networks. However, one of the problems encountered in the training of very deep networks is feature reuse; that is, features are ’diluted’ as they are forward propagated through the model. Hence, later network layers receive less informative signals about the input data, consequently making training less effective. In this work, we address the problem of feature reuse by taking inspiration from an earlier work which employed residual learning for alleviating the problem of feature reuse. We propose a modification of residual learning for training very deep networks to realize improved generalization performance; for this, we allow stochastic shortcut connections of identity mappings from the input to hidden layers.We perform extensive experiments using the USPS and MNIST datasets. On the USPS dataset, we achieve an error rate of 2.69% without employing any form of data augmentation (or manipulation). On the MNIST dataset, we reach a comparable state-of-the-art error rate of 0.52%. Particularly, these results are achieved without employing any explicit regularization technique. [less ▲]

Detailed reference viewed: 269 (46 UL)