There exists many resource allocation problems in the field of wireless communications which can be formulated as the generalized assignment problems (GAP). GAP is a generic form of linear sum assignment problem (LSAP) and is more challenging to solve owing to the presence of both equality and inequality constraints. We propose a novel deep unsupervised learning (DUL) approach to solve GAP in a time-efficient manner. More specifically, we propose a new approach that facilitates to train a deep neural network (DNN) using a customized loss function. This customized loss function constitutes the objective function and penalty terms corresponding to both equality and inequality constraints. Furthermore, we propose to employ a Softmax activation function at the output of DNN along with tensor splitting which simplifies the customized loss function and guarantees to meet the equality constraint. As a case-study, we consider a typical user-association problem in a wireless network, formulate it as GAP, and consequently solve it using our proposed DUL approach. Numerical results demonstrate that the proposed DUL approach provides near-optimal results with significantly lower time-complexity.
In video transformers, the time dimension is often treated in the same way as the two spatial dimensions. However, in a scene where objects or the camera may move, a physical point imaged at one location in frame $t$ may be entirely unrelated to what is found at that location in frame $t+k$. These temporal correspondences should be modeled to facilitate learning about dynamic scenes. To this end, we propose a new drop-in block for video transformers -- trajectory attention -- that aggregates information along implicitly determined motion paths. We additionally propose a new method to address the quadratic dependence of computation and memory on the input size, which is particularly important for high resolution or long videos. While these ideas are useful in a range of settings, we apply them to the specific task of video action recognition with a transformer model and obtain state-of-the-art results on the Kinetics, Something--Something V2, and Epic-Kitchens datasets. Code and models are available at: https://github.com/facebookresearch/Motionformer
We address scaling up equilibrium computation in Mean Field Games (MFGs) using Online Mirror Descent (OMD). We show that continuous-time OMD provably converges to a Nash equilibrium under a natural and well-motivated set of monotonicity assumptions. This theoretical result nicely extends to multi-population games and to settings involving common noise. A thorough experimental investigation on various single and multi-population MFGs shows that OMD outperforms traditional algorithms such as Fictitious Play (FP). We empirically show that OMD scales up and converges significantly faster than FP by solving, for the first time to our knowledge, examples of MFGs with hundreds of billions states. This study establishes the state-of-the-art for learning in large-scale multi-agent and multi-population games.
High screening coverage during colonoscopy is crucial to effectively prevent colon cancer. Previous work has allowed alerting the doctor to unsurveyed regions by reconstructing the 3D colonoscopic surface from colonoscopy videos in real-time. However, the lighting inconsistency of colonoscopy videos can cause a key component of the colonoscopic reconstruction system, the SLAM optimization, to fail. In this work we focus on the lighting problem in colonoscopy videos. To successfully improve the lighting consistency of colonoscopy videos, we have found necessary a lighting correction that adapts to the intensity distribution of recent video frames. To achieve this in real-time, we have designed and trained an RNN network. This network adapts the gamma value in a gamma-correction process. Applied in the colonoscopic surface reconstruction system, our light-weight model significantly boosts the reconstruction success rate, making a larger proportion of colonoscopy video segments reconstructable and improving the reconstruction quality of the already reconstructed segments.
Enforcing orthogonality in neural networks is an antidote for gradient vanishing/exploding problems, sensitivity by adversarial perturbation, and bounding generalization errors. However, many previous approaches are heuristic, and the orthogonality of convolutional layers is not systematically studied: some of these designs are not exactly orthogonal, while others only consider standard convolutional layers and propose specific classes of their realizations. To address this problem, we propose a theoretical framework for orthogonal convolutional layers, which establishes the equivalence between various orthogonal convolutional layers in the spatial domain and the paraunitary systems in the spectral domain. Since there exists a complete spectral factorization of paraunitary systems, any orthogonal convolution layer can be parameterized as convolutions of spatial filters. Our framework endows high expressive power to various convolutional layers while maintaining their exact orthogonality. Furthermore, our layers are memory and computationally efficient for deep networks compared to previous designs. Our versatile framework, for the first time, enables the study of architecture designs for deep orthogonal networks, such as choices of skip connection, initialization, stride, and dilation. Consequently, we scale up orthogonal networks to deep architectures, including ResNet, WideResNet, and ShuffleNet, substantially increasing the performance over the traditional shallow orthogonal networks.
In this paper, we propose a simple methodology to detect the partial pose of a human occupying the manipulator work-space using only on-robot time--of--flight laser ranging sensors. The sensors are affixed on each link of the robot in a circular array fashion where each array possesses sixteen single unit laser ranging lidar(s). The detection is performed by leveraging an artificial neural network which takes a highly sparse 3-D point cloud input to produce an estimate of the partial pose which is the ground projection frame of the human footprint. We also present a particle filter based approach to the tracking problem when the input data is unreliable. Ultimately, the simulation results are presented and analyzed.
The vast majority of semantic segmentation approaches rely on pixel-level annotations that are tedious and time consuming to obtain and suffer from significant inter and intra-expert variability. To address these issues, recent approaches have leveraged categorical annotations at the slide-level, that in general suffer from robustness and generalization. In this paper, we propose a novel weakly supervised multi-instance learning approach that deciphers quantitative slide-level annotations which are fast to obtain and regularly present in clinical routine. The extreme potentials of the proposed approach are demonstrated for tumor segmentation of solid cancer subtypes. The proposed approach achieves superior performance in out-of-distribution, out-of-location, and out-of-domain testing sets.
Processing point cloud data is an important component of many real-world systems. As such, a wide variety of point-based approaches have been proposed, reporting steady benchmark improvements over time. We study the key ingredients of this progress and uncover two critical results. First, we find that auxiliary factors like different evaluation schemes, data augmentation strategies, and loss functions, which are independent of the model architecture, make a large difference in performance. The differences are large enough that they obscure the effect of architecture. When these factors are controlled for, PointNet++, a relatively older network, performs competitively with recent methods. Second, a very simple projection-based method, which we refer to as SimpleView, performs surprisingly well. It achieves on par or better results than sophisticated state-of-the-art methods on ModelNet40 while being half the size of PointNet++. It also outperforms state-of-the-art methods on ScanObjectNN, a real-world point cloud benchmark, and demonstrates better cross-dataset generalization. Code is available at https://github.com/princeton-vl/SimpleView.
The new generation of pre-trained NLP models push the SOTA to the new limits, but at the cost of computational resources, to the point that their use in real production environments is often prohibitively expensive. We tackle this problem by evaluating not only the standard quality metrics on downstream tasks but also the memory footprint and inference time. We present MOROCCO, a framework to compare language models compatible with \texttt{jiant} environment which supports over 50 NLU tasks, including SuperGLUE benchmark and multiple probing suites. We demonstrate its applicability for two GLUE-like suites in different languages.
Modelling mix-and-match relationships among fashion items has become increasingly demanding yet challenging for modern E-commerce recommender systems. When performing clothes matching, most existing approaches leverage the latent visual features extracted from fashion item images for compatibility modelling, which lacks explainability of generated matching results and can hardly convince users of the recommendations. Though recent methods start to incorporate pre-defined attribute information (e.g., colour, style, length, etc.) for learning item representations and improving the model interpretability, their utilisation of attribute information is still mainly reserved for enhancing the learned item representations and generating explanations via post-processing. As a result, this creates a severe bottleneck when we are trying to advance the recommendation accuracy and generating fine-grained explanations since the explicit attributes have only loose connections to the actual recommendation process. This work aims to tackle the explainability challenge in fashion recommendation tasks by proposing a novel Attribute-aware Fashion Recommender (AFRec). Specifically, AFRec recommender assesses the outfit compatibility by explicitly leveraging the extracted attribute-level representations from each item's visual feature. The attributes serve as the bridge between two fashion items, where we quantify the affinity of a pair of items through the learned compatibility between their attributes. Extensive experiments have demonstrated that, by making full use of the explicit attributes in the recommendation process, AFRec is able to achieve state-of-the-art recommendation accuracy and generate intuitive explanations at the same time.