arXiv subjects

Chen-Yu Lee

Publications and source records attributed to Chen-Yu Lee.

66 records · Page 4Linked to original sources

arXiv2020

A Simple Semi-Supervised Learning Framework for Object Detection

Semi-supervised learning (SSL) has a potential to improve the predictive performance of machine learning models using unlabeled data. Although there has been remarkable recent progress, the scope of demonstration in SSL has mainly been on image classification tasks. In this paper, we propose STAC, a simple yet effective SSL framework for visual object detection along with a data augmentation strategy. STAC deploys highly confident pseudo labels of localized objects from an unlabeled image and updates the model by enforcing consistency via strong augmentations. We propose experimental protocols to evaluate the performance of semi-supervised object detection using MS-COCO and show the efficacy of STAC on both MS-COCO and VOC07. On VOC07, STAC improves the AP$^{0.5}$ from $76.30$ to $79.08$; on MS-COCO, STAC demonstrates $2{\times}$ higher data efficiency by achieving 24.38 mAP using only 5\% labeled data than supervised baseline that marks 23.86\% using 10\% labeled data. The code is available at https://github.com/google-research/ssl_detection/.

cs.CV↗

arXiv2020

Prototype of a phonon laser with trapped ions

We propose a tunable phonon laser prototype with a large trapped ion array, where some of the ions are effectively pinned by optical tweezers, thus isolating a subset of ions that mimics an acoustic cavity used as a phonon lasing resonator. The cavity loss can then be controlled by the tweezer strength and the "wall thickness", the number of pinned ions for isolation. We pump the resonator by applying blue-sideband lasers, and investigate the lasing dynamics of the cavity modes such as threshold behavior, population distribution, the second-order coherence, and line-narrowed spectrum. This scheme can be generalized to resonators consisting of multiple cavity modes formed by a few ions, where we demonstrate mode competition and synchronization as lasing modes have developed.

quant-ph↗

arXiv2020

Learning to Branch for Multi-Task Learning

Training multiple tasks jointly in one deep network yields reduced latency during inference and better performance over the single-task counterpart by sharing certain layers of a network. However, over-sharing a network could erroneously enforce over-generalization, causing negative knowledge transfer across tasks. Prior works rely on human intuition or pre-computed task relatedness scores for ad hoc branching structures. They provide sub-optimal end results and often require huge efforts for the trial-and-error process. In this work, we present an automated multi-task learning algorithm that learns where to share or branch within a network, designing an effective network topology that is directly optimized for multiple objectives across tasks. Specifically, we propose a novel tree-structured design space that casts a tree branching operation as a gumbel-softmax sampling procedure. This enables differentiable network splitting that is end-to-end trainable. We validate the proposed method on controlled synthetic data, CelebA, and Taskonomy.

cs.LG↗

arXiv2019

Scalable collective Lamb shift of a 1D superconducting qubit array in front of a mirror

We theoretically investigate resonant dipole-dipole interaction (RDDI) between artificial atoms in a 1D geometry, implemented by N transmon qubits coupled through a transmission line. Similarly to the atomic cases, RDDI comes from exchange of virtual photons of the unexcited modes, and causes the so-called collective Lamb shift (CLS). To probe the shift, we effectively set one end of the transmission line as a mirror, and examine the reflection spectrum of the probe field from the other end. Our calculation shows that when a qubit is placed at the node of the standing wave formed by the incident and reflected waves, even though it is considered to be decoupled from the field, it results in large energy splitting in the spectral profile of a resonant qubit located elsewhere. This directly signals the interplay of virtual photon processes and explicitly demonstrates the CLS. We further derive a master equation to describe the system, which can take into account mismatch of participating qubits and dephasing effects. Our calculation also demonstrates the superradiant and subradiant nature of the atomic states, and how the CLS scales when more qubits are involved.

quant-ph↗

arXiv2019

Sliced Wasserstein Discrepancy for Unsupervised Domain Adaptation

In this work, we connect two distinct concepts for unsupervised domain adaptation: feature distribution alignment between domains by utilizing the task-specific decision boundary and the Wasserstein metric. Our proposed sliced Wasserstein discrepancy (SWD) is designed to capture the natural notion of dissimilarity between the outputs of task-specific classifiers. It provides a geometrically meaningful guidance to detect target samples that are far from the support of the source and enables efficient distribution alignment in an end-to-end trainable fashion. In the experiments, we validate the effectiveness and genericness of our method on digit and sign recognition, image classification, semantic segmentation, and object detection.

cs.CV↗

arXiv2018

GradNorm: Gradient Normalization for Adaptive Loss Balancing in Deep Multitask Networks

Deep multitask networks, in which one neural network produces multiple predictive outputs, can offer better speed and performance than their single-task counterparts but are challenging to train properly. We present a gradient normalization (GradNorm) algorithm that automatically balances training in deep multitask models by dynamically tuning gradient magnitudes. We show that for various network architectures, for both regression and classification tasks, and on both synthetic and real datasets, GradNorm improves accuracy and reduces overfitting across multiple tasks when compared to single-task networks, static baselines, and other adaptive multitask loss balancing techniques. GradNorm also matches or surpasses the performance of exhaustive grid search methods, despite only involving a single asymmetry hyperparameter $α$. Thus, what was once a tedious search process that incurred exponentially more compute for each task added can now be accomplished within a few training runs, irrespective of the number of tasks. Ultimately, we will demonstrate that gradient manipulation affords us great control over the training dynamics of multitask networks and may be one of the keys to unlocking the potential of multitask learning.

cs.CV↗

arXiv2017

RoomNet: End-to-End Room Layout Estimation

This paper focuses on the task of room layout estimation from a monocular RGB image. Prior works break the problem into two sub-tasks: semantic segmentation of floor, walls, ceiling to produce layout hypotheses, followed by an iterative optimization step to rank these hypotheses. In contrast, we adopt a more direct formulation of this problem as one of estimating an ordered set of room layout keypoints. The room layout and the corresponding segmentation is completely specified given the locations of these ordered keypoints. We predict the locations of the room layout keypoints using RoomNet, an end-to-end trainable encoder-decoder network. On the challenging benchmark datasets Hedau and LSUN, we achieve state-of-the-art performance along with 200x to 600x speedup compared to the most recent work. Additionally, we present optional extensions to the RoomNet architecture such as including recurrent computations and memory units to refine the keypoint locations under the same parametric capacity.

cs.CV↗

arXiv2016

Recursive Recurrent Nets with Attention Modeling for OCR in the Wild

We present recursive recurrent neural networks with attention modeling (R$^2$AM) for lexicon-free optical character recognition in natural scene images. The primary advantages of the proposed method are: (1) use of recursive convolutional neural networks (CNNs), which allow for parametrically efficient and effective image feature extraction; (2) an implicitly learned character-level language model, embodied in a recurrent neural network which avoids the need to use N-grams; and (3) the use of a soft-attention mechanism, allowing the model to selectively exploit image features in a coordinated way, and allowing for end-to-end training within a standard backpropagation framework. We validate our method with state-of-the-art performance on challenging benchmark datasets: Street View Text, IIIT5k, ICDAR and Synth90k.

cs.CV↗

arXiv2015

Generalizing Pooling Functions in Convolutional Neural Networks: Mixed, Gated, and Tree

We seek to improve deep neural networks by generalizing the pooling operations that play a central role in current architectures. We pursue a careful exploration of approaches to allow pooling to learn and to adapt to complex and variable patterns. The two primary directions lie in (1) learning a pooling function via (two strategies of) combining of max and average pooling, and (2) learning a pooling function in the form of a tree-structured fusion of pooling filters that are themselves learned. In our experiments every generalized pooling operation we explore improves performance when used in place of average or max pooling. We experimentally demonstrate that the proposed pooling operations provide a boost in invariance properties relative to conventional pooling and set the state of the art on several widely adopted benchmark datasets; they are also easy to implement, and can be applied within various deep neural network architectures. These benefits come with only a light increase in computational overhead during training and a very modest increase in the number of model parameters.

stat.ML↗

arXiv2015

Training Deeper Convolutional Networks with Deep Supervision

One of the most promising ways of improving the performance of deep convolutional neural networks is by increasing the number of convolutional layers. However, adding layers makes training more difficult and computationally expensive. In order to train deeper networks, we propose to add auxiliary supervision branches after certain intermediate layers during training. We formulate a simple rule of thumb to determine where these branches should be added. The resulting deeply supervised structure makes the training much easier and also produces better classification results on ImageNet and the recently released, larger MIT Places dataset

cs.CV↗

arXiv2015

A Content Creation and Protection Scheme for Medical Images

Medical images contain metadata information on where, when, and how an image was acquired, and the majority of this information is stored as pixel data. Image feature descriptions are often captured only as free text stored in the image file or in the hospital information system. Correlations between the free text and the location of the feature are often inaccurate, making it difficult to link image observations to their corresponding image locations. This limits the interpretation of image data from a clinical, research, and academic standpoint. An efficient medical image protection design should allow for compatibility, usability, and privacy. This paper proposes a medical-content creation and protection scheme that contains a) a DICOM-compatible multimedia annotation scheme for medical content creation; b) a DICOM-compatible partial DRM scheme for medical record transmission under this scheme, authorized users can view only information to which they have been granted to access.

cs.CR↗

arXiv2014

Deeply-Supervised Nets

Our proposed deeply-supervised nets (DSN) method simultaneously minimizes classification error while making the learning process of hidden layers direct and transparent. We make an attempt to boost the classification performance by studying a new formulation in deep networks. Three aspects in convolutional neural networks (CNN) style architectures are being looked at: (1) transparency of the intermediate layers to the overall classification; (2) discriminativeness and robustness of learned features, especially in the early layers; (3) effectiveness in training due to the presence of the exploding and vanishing gradients. We introduce "companion objective" to the individual hidden layers, in addition to the overall objective at the output layer (a different strategy to layer-wise pre-training). We extend techniques from stochastic gradient methods to analyze our algorithm. The advantage of our method is evident and our experimental result on benchmark datasets shows significant performance gain over existing methods (e.g. all state-of-the-art results on MNIST, CIFAR-10, CIFAR-100, and SVHN).

stat.ML↗