arXiv subjects

Boris Chidlovskii

Publications and source records attributed to Boris Chidlovskii.

28 records · Page 2Linked to original sources

arXiv2021

Learning From Long-Tailed Data With Noisy Labels

Class imbalance and noisy labels are the norm rather than the exception in many large-scale classification datasets. Nevertheless, most works in machine learning typically assume balanced and clean data. There have been some recent attempts to tackle, on one side, the problem of learning from noisy labels and, on the other side, learning from long-tailed data. Each group of methods make simplifying assumptions about the other. Due to this separation, the proposed solutions often underperform when both assumptions are violated. In this work, we present a simple two-stage approach based on recent advances in self-supervised learning to treat both challenges simultaneously. It consists of, first, task-agnostic self-supervised pre-training, followed by task-specific fine-tuning using an appropriate loss. Most significantly, we find that self-supervised learning approaches are effectively able to cope with severe class imbalance. In addition, the resulting learned representations are also remarkably robust to label noise, when fine-tuned with an imbalance- and noise-resistant loss function. We validate our claims with experiments on CIFAR-10 and CIFAR-100 augmented with synthetic imbalance and noise, as well as the large-scale inherently noisy Clothing-1M dataset.

cs.CV↗

arXiv2021

Magnetic Field Sensing for Pedestrian and Robot Indoor Positioning

In this paper we address the problem of indoor localization using magnetic field data in two setups, when data is collected by (i) human-held mobile phone and (ii) by localization robots that perturb magnetic data with their own electromagnetic field. For the first setup, we revise the state of the art approaches and propose a novel extended pipeline to benefit from the presence of magnetic anomalies in indoor environment created by different ferromagnetic objects. We capture changes of the Earth's magnetic field due to indoor magnetic anomalies and transform them in multi-variate times series. We then convert temporal patterns into visual ones. We use methods of Recurrence Plots, Gramian Angular Fields and Markov Transition Fields to represent magnetic field time series as image sequences. We regress the continuous values of user position in a deep neural network that combines convolutional and recurrent layers. For the second setup, we analyze how magnetic field data get perturbed by robots' electromagnetic field. We add an alignment step to the main pipeline, in order to compensate the mismatch between train and test sets obtained by different robots. We test our methods on two public (MagPie and IPIN'20) and one proprietary (Hyundai department store) datasets. We report evaluation results and show that our methods outperform the state of the art methods by a large margin.

cs.RO↗

arXiv2021

Universal Domain Adaptation in Ordinal Regression

We address the problem of universal domain adaptation (UDA) in ordinal regression (OR), which attempts to solve classification problems in which labels are not independent, but follow a natural order. We show that the UDA techniques developed for classification and based on the clustering assumption, under-perform in OR settings. We propose a method that complements the OR classifier with an auxiliary task of order learning, which plays the double role of discriminating between common and private instances, and expanding class labels to the private target images via ranking. Combined with adversarial domain discrimination, our model is able to address the closed set, partial and open set configurations. We evaluate our method on three face age estimation datasets, and show that it outperforms the baseline methods.

cs.CV↗

arXiv2020

Adversarial Transfer of Pose Estimation Regression

We address the problem of camera pose estimation in visual localization. Current regression-based methods for pose estimation are trained and evaluated scene-wise. They depend on the coordinate frame of the training dataset and show a low generalization across scenes and datasets. We identify the dataset shift an important barrier to generalization and consider transfer learning as an alternative way towards a better reuse of pose estimation models. We revise domain adaptation techniques for classification and extend them to camera pose estimation, which is a multi-regression task. We develop a deep adaptation network for learning scene-invariant image representations and use adversarial learning to generate such representations for model transfer. We enrich the network with self-supervised learning and use the adaptability theory to validate the existence of scene-invariant representation of images in two given scenes. We evaluate our network on two public datasets, Cambridge Landmarks and 7Scene, demonstrate its superiority over several baselines and compare to the state of the art methods.

cs.CV↗

arXiv2020

Learning Synthetic to Real Transfer for Localization and Navigational Tasks

Autonomous navigation consists in an agent being able to navigate without human intervention or supervision, it affects both high level planning and low level control. Navigation is at the crossroad of multiple disciplines, it combines notions of computer vision, robotics and control. This work aimed at creating, in a simulation, a navigation pipeline whose transfer to the real world could be done with as few efforts as possible. Given the limited time and the wide range of problematic to be tackled, absolute navigation performances while important was not the main objective. The emphasis was rather put on studying the sim2real gap which is one the major bottlenecks of modern robotics and autonomous navigation. To design the navigation pipeline four main challenges arise; environment, localization, navigation and planning. The iGibson simulator is picked for its photo-realistic textures and physics engine. A topological approach to tackle space representation was picked over metric approaches because they generalize better to new environments and are less sensitive to change of conditions. The navigation pipeline is decomposed as a localization module, a planning module and a local navigation module. These modules utilize three different networks, an image representation extractor, a passage detector and a local policy. The laters are trained on specifically tailored tasks with some associated datasets created for those specific tasks. Localization is the ability for the agent to localize itself against a specific space representation. It must be reliable, repeatable and robust to a wide variety of transformations. Localization is tackled as an image retrieval task using a deep neural network trained on an auxiliary task as a feature descriptor extractor. The local policy is trained with behavioral cloning from expert trajectories gathered with ROS navigation stack.

cs.RO↗

arXiv2020

Deep Smartphone Sensors-WiFi Fusion for Indoor Positioning and Tracking

We address the indoor localization problem, where the goal is to predict user's trajectory from the data collected by their smartphone, using inertial sensors such as accelerometer, gyroscope and magnetometer, as well as other environment and network sensors such as barometer and WiFi. Our system implements a deep learning based pedestrian dead reckoning (deep PDR) model that provides a high-rate estimation of the relative position of the user. Using Kalman Filter, we correct the PDR's drift using WiFi that provides a prediction of the user's absolute position each time a WiFi scan is received. Finally, we adjust Kalman Filter results with a map-free projection method that takes into account the physical constraints of the environment (corridors, doors, etc.) and projects the prediction on the possible walkable paths. We test our pipeline on IPIN'19 Indoor Localization challenge dataset and demonstrate that it improves the winner's results by 20\% using the challenge evaluation protocol.

cs.LG↗

arXiv2019

Using Latent Codes for Class Imbalance Problem in Unsupervised Domain Adaptation

We address the problem of severe class imbalance in unsupervised domain adaptation, when the class spaces in source and target domains diverge considerably. Till recently, domain adaptation methods assumed the aligned class spaces, such that reducing distribution divergence makes the transfer between domains easier. Such an alignment assumption is invalidated in real world scenarios where some source classes are often under-represented or simply absent in the target domain. We revise the current approaches to class imbalance and propose a new one that uses latent codes in the adversarial domain adaptation framework. We show how the latent codes can be used to disentangle the silent structure of the target domain and to identify under-represented classes. We show how to learn the latent code reconstruction jointly with the domain invariant representation and use them to accurately estimate the target labels.

cs.LG↗

arXiv2018

Learning Common Representation from RGB and Depth Images

We propose a new deep learning architecture for the tasks of semantic segmentation and depth prediction from RGB-D images. We revise the state of art based on the RGB and depth feature fusion, where both modalities are assumed to be available at train and test time. We propose a new architecture where the feature fusion is replaced with a common deep representation. Combined with an encoder-decoder type of the network, the architecture can jointly learn models for semantic segmentation and depth estimation based on their common representation. This representation, inspired by multi-view learning, offers several important advantages, such as using one modality available at test time to reconstruct the missing modality. In the RGB-D case, this enables the cross-modality scenarios, such as using depth data for semantically segmentation and the RGB images for depth estimation. We demonstrate the effectiveness of the proposed network on two publicly available RGB-D datasets. The experimental results show that the proposed method works well in both semantic segmentation and depth estimation tasks.

cs.AI↗

arXiv2017

Multi-task learning of time series and its application to the travel demand

We address the problem of modeling and prediction of a set of temporal events in the context of intelligent transportation systems. To leverage the information shared by different events, we propose a multi-task learning framework. We develop a support vector regression model for joint learning of mutually dependent time series. It is the regularization-based multi-task learning previously developed for the classification case and extended to time series. We discuss the relatedness of observed time series and first deploy the dynamic time warping distance measure to identify groups of similar series. Then we take into account both time and scale warping and propose to align multiple time series by inferring their common latent representation. We test the proposed models on the problem of travel demand prediction in Nancy (France) public transport system and analyze the benefits of multi-task learning.

cs.LG↗

arXiv2017

Mining Smart Card Data for Travelers' Mini Activities

In the context of public transport modeling and simulation, we address the problem of mismatch between simulated transit trips and observed ones. We point to the weakness of the current travel demand modeling process; the trips it generates are over-optimistic and do not reflect the real passenger choices. We introduce the notion of mini activities the travelers do during the trips; they can explain the deviation of simulated trips from the observed trips. We propose to mine the smart card data to extract the mini activities. We develop a technique to integrate them in the generated trips and learn such an integration from two available sources, the trip history and trip planner recommendations. For an input travel demand, we build a Markov chain over the trip collection and apply the Monte Carlo Markov Chain algorithm to integrate mini activities in such a way that the selected characteristics converge to the desired distributions. We test our method in different settings on the passenger trip collection of Nancy, France. We report experimental results demonstrating a very important mismatch reduction.

cs.AI↗