There are many ways to represent a molecule as input to a machine learning model and each is associated with loss and retention of certain kinds of information. In the interest of preserving three-dimensional spatial information, including bond angles and torsions, we have developed libmolgrid, a general-purpose library for representing three-dimensional molecules using multidimensional arrays. This library also provides functionality for composing batches of data suited to machine learning workflows, including data augmentation, class balancing, and example stratification according to a regression variable or data subgroup, and it further supports temporal and spatial recurrences over that data to facilitate work with recurrent neural networks, dynamical data, and size extensive modeling. It was designed for seamless integration with popular deep learning frameworks, including Caffe, PyTorch, and Keras, providing good performance by leveraging graphical processing units (GPUs) for computationally-intensive tasks and efficient memory usage through the use of memory views over preallocated buffers. libmolgrid is a free and open source project that is actively supported, serving the growing need in the molecular modeling community for tools that streamline the process of data ingestion, representation construction, and principled machine learning model development.
Over the last years, a great success of deep neural networks (DNNs) has been witnessed in computer vision and other fields. However, performance and power constraints make it still challenging to deploy DNNs on mobile devices due to their high computational complexity. Binary neural networks (BNNs) have been demonstrated as a promising solution to achieve this goal by using bit-wise operations to replace most arithmetic operations. Currently, existing GPU-accelerated implementations of BNNs are only tailored for desktop platforms. Due to architecture differences, mere porting of such implementations to mobile devices yields suboptimal performance or is impossible in some cases. In this paper, we propose PhoneBit, a GPU-accelerated BNN inference engine for Android-based mobile devices that fully exploits the computing power of BNNs on mobile GPUs. PhoneBit provides a set of operator-level optimizations including locality-friendly data layout, bit packing with vectorization and layers integration for efficient binary convolution. We also provide a detailed implementation and parallelization optimization for PhoneBit to optimally utilize the memory bandwidth and computing power of mobile GPUs. We evaluate PhoneBit with AlexNet, YOLOv2 Tiny and VGG16 with their binary version. Our experiment results show that PhoneBit can achieve significant speedup and energy efficiency compared with state-of-the-art frameworks for mobile devices.
This work presents Origami, which provides privacy-preserving inference for large deep neural network (DNN) models through a combination of enclave execution, cryptographic blinding, interspersed with accelerator-based computation. Origami partitions the ML model into multiple partitions. The first partition receives the encrypted user input within an SGX enclave. The enclave decrypts the input and then applies cryptographic blinding to the input data and the model parameters. Cryptographic blinding is a technique that adds noise to obfuscate data. Origami sends the obfuscated data for computation to an untrusted GPU/CPU. The blinding and de-blinding factors are kept private by the SGX enclave, thereby preventing any adversary from denoising the data, when the computation is offloaded to a GPU/CPU. The computed output is returned to the enclave, which decodes the computation on noisy data using the unblinding factors privately stored within SGX. This process may be repeated for each DNN layer, as has been done in prior work Slalom. However, the overhead of blinding and unblinding the data is a limiting factor to scalability. Origami relies on the empirical observation that the feature maps after the first several layers can not be used, even by a powerful conditional GAN adversary to reconstruct input. Hence, Origami dynamically switches to executing the rest of the DNN layers directly on an accelerator without needing any further cryptographic blinding intervention to preserve privacy. We empirically demonstrate that using Origami, a conditional GAN adversary, even with an unlimited inference budget, cannot reconstruct the input. We implement and demonstrate the performance gains of Origami using the VGG-16 and VGG-19 models. Compared to running the entire VGG-19 model within SGX, Origami inference improves the performance of private inference from 11x while using Slalom to 15.1x.
We introduce JAX MD, a software package for performing differentiable physics simulations with a focus on molecular dynamics. JAX MD includes a number of physics simulation environments, as well as interaction potentials and neural networks that can be integrated into these environments without writing any additional code. Since the simulations themselves are differentiable functions, entire trajectories can be differentiated to perform meta-optimization. These features are built on primitive operations, such as spatial partitioning, that allow simulations to scale to hundreds-of-thousands of particles on a single GPU. These primitives are flexible enough that they can be used to scale up workloads outside of molecular dynamics. We present several examples that highlight the features of JAX MD including: integration of graph neural networks into traditional simulations, meta-optimization through minimization of particle packings, and a multi-agent flocking simulation. JAX MD is available at www.github.com/google/jax-md.
Neural Tangents is a library designed to enable research into infinite-width neural networks. It provides a high-level API for specifying complex and hierarchical neural network architectures. These networks can then be trained and evaluated either at finite-width as usual or in their infinite-width limit. Infinite-width networks can be trained analytically using exact Bayesian inference or using gradient descent via the Neural Tangent Kernel. Additionally, Neural Tangents provides tools to study gradient descent training dynamics of wide but finite networks in either function space or weight space. The entire library runs out-of-the-box on CPU, GPU, or TPU. All computations can be automatically distributed over multiple accelerators with near-linear scaling in the number of devices. Neural Tangents is available at www.github.com/google/neural-tangents. We also provide an accompanying interactive Colab notebook.
The objective of ordinal embedding is to find a Euclidean representation of a set of abstract items, using only answers to triplet comparisons of the form "Is item $i$ closer to the item $j$ or item $k$?". In recent years, numerous algorithms have been proposed to solve this problem. However, there does not exist a fair and thorough assessment of these embedding methods and therefore several key questions remain unanswered: Which algorithms perform better when the embedding dimension is constrained or few triplet comparisons are available? Which ones scale better with increasing sample size or dimension? In our paper, we address these questions and provide the first comprehensive and systematic empirical evaluation of existing algorithms as well as a new neural network approach. We find that simple, relatively unknown, non-convex methods consistently outperform all other algorithms, including elaborate approaches based on neural networks or landmark approaches. This finding can be explained by our insight that many of the non-convex optimization approaches do not suffer from local optima. Our comprehensive assessment is enabled by our unified library of popular embedding algorithms that leverages GPU resources and allows for fast and accurate embeddings of millions of data points.
We propose a novel Siamese Natural Language Tracker (SNLT), which brings the advancements in visual tracking to the tracking by natural language (NL) descriptions task. The proposed SNLT is applicable to a wide range of Siamese trackers, providing a new class of baselines for the tracking by NL task and promising future improvements from the advancements of Siamese trackers. The carefully designed architecture of the Siamese Natural Language Region Proposal Network (SNL-RPN), together with the Dynamic Aggregation of vision and language modalities, is introduced to perform the tracking by NL task. Empirical results over tracking benchmarks with NL annotations show that the proposed SNLT improves Siamese trackers by 3 to 7 percentage points with a slight tradeoff of speed. The proposed SNLT outperforms all NL trackers to-date and is competitive among state-of-the-art real-time trackers on LaSOT benchmarks while running at 50 frames per second on a single GPU.
Neural Architecture Search (NAS) that aims to automate the procedure of architecture design has achieved promising results in many computer vision fields. In this paper, we propose an AdversarialNAS method specially tailored for Generative Adversarial Networks (GANs) to search for a superior generative model on the task of unconditional image generation. The AdversarialNAS is the first method that can search the architectures of generator and discriminator simultaneously in a differentiable manner. During searching, the designed adversarial search algorithm does not need to comput any extra metric to evaluate the performance of the searched architecture, and the search paradigm considers the relevance between the two network architectures and improves their mutual balance. Therefore, AdversarialNAS is very efficient and only takes 1 GPU day to search for a superior generative model in the proposed large search space ($10^{38}$). Experiments demonstrate the effectiveness and superiority of our method. The discovered generative model sets a new state-of-the-art FID score of $10.87$ and highly competitive Inception Score of $8.74$ on CIFAR-10. Its transferability is also proven by setting new state-of-the-art FID score of $26.98$ and Inception score of $9.63$ on STL-10. Code is at: \url{https://github.com/chengaopro/AdversarialNAS}.
In this work, we propose WaveFlow, a small-footprint generative flow for raw audio, which is directly trained with maximum likelihood. It handles the long-range structure of 1-D waveform with a dilated 2-D convolutional architecture, while modeling the local variations using expressive autoregressive functions. WaveFlow provides a unified view of likelihood-based models for 1-D data, including WaveNet and WaveGlow as special cases. It generates high-fidelity speech as WaveNet, while synthesizing several orders of magnitude faster as it only requires a few sequential steps to generate very long waveforms with hundreds of thousands of time-steps. Furthermore, it can significantly reduce the likelihood gap that has existed between autoregressive models and flow-based models for efficient synthesis. Finally, our small-footprint WaveFlow has only 5.91M parameters, which is 15$\times$ smaller than WaveGlow. It can generate 22.05 kHz high-fidelity audio 42.6$\times$ faster than real-time (at a rate of 939.3 kHz) on a V100 GPU without engineered inference kernels.
Transferrable neural architecture search can be viewed as a binary optimization problem where a single optimal path should be selected among candidate paths in each edge within the repeated cell block of the directed a cyclic graph form. Recently, the field of differentiable architecture search attempts to relax the search problem continuously using a one-shot network that combines all the candidate paths in search space. However, when the one-shot network is pruned to the model in the discrete architecture space by the derivation algorithm, performance is significantly degraded to an almost random estimator. To reduce the quantization error from the heavy use of relaxation, we only sample a single edge to relax the corresponding variable and clamp variables in the other edges to zero or one. By this method, there is no performance drop after pruning the one-shot network by derivation algorithm, due to the preservation of the discrete nature of optimization variables during the search. Furthermore, the minimization of relaxation degree allows searching in a deeper network to discover better performance with remarkable search cost reduction (0.125 GPU days) compared to previous methods. By adding several regularization methods that help explore within the search space, we could obtain the network with notable performances on CIFAR-10, CIFAR-100, and ImageNet.
Approximate nearest neighbor (ANN) search in high dimensions is an integral part of several computer vision systems and gains importance in deep learning with explicit memory representations. Since PQT, FAISS, and SONG started to leverage the massive parallelism offered by GPUs, GPU-based implementations are a crucial resource for today's state-of-the-art ANN methods. While most of these methods allow for faster queries, less emphasis is devoted to accelerating the construction of the underlying index structures. In this paper, we propose a novel GPU-friendly search structure based on nearest neighbor graphs and information propagation on graphs. Our method is designed to take advantage of GPU architectures to accelerate the hierarchical construction of the index structure and for performing the query. Empirical evaluation shows that GGNN significantly surpasses the state-of-the-art CPU- and GPU-based systems in terms of build-time, accuracy and search speed.
Training competitive deep video models is an order of magnitude slower than training their counterpart image models. Slow training causes long research cycles, which hinders progress in video understanding research. Following standard practice for training image models, video model training assumes a fixed mini-batch shape: a specific number of clips, frames, and spatial size. However, what is the optimal shape? High resolution models perform well, but train slowly. Low resolution models train faster, but they are inaccurate. Inspired by multigrid methods in numerical optimization, we propose to use variable mini-batch shapes with different spatial-temporal resolutions that are varied according to a schedule. The different shapes arise from resampling the training data on multiple sampling grids. Training is accelerated by scaling up the mini-batch size and learning rate when shrinking the other dimensions. We empirically demonstrate a general and robust grid schedule that yields a significant out-of-the-box training speedup without a loss in accuracy for different models (I3D, non-local, SlowFast), datasets (Kinetics, Something-Something, Charades), and training settings (with and without pre-training, 128 GPUs or 1 GPU). As an illustrative example, the proposed multigrid method trains a ResNet-50 SlowFast network 4.5x faster (wall-clock time, same hardware) while also improving accuracy (+0.8% absolute) on Kinetics-400 compared to the baseline training method. Code is available online.
Current state-of-the-art models for video action recognition are mostly based on expensive 3D ConvNets. This results in a need for large GPU clusters to train and evaluate such architectures. To address this problem, we present a lightweight and memory-friendly architecture for action recognition that performs on par with or better than current architectures by using only a fraction of resources. The proposed architecture is based on a combination of a deep subnet operating on low-resolution frames with a compact subnet operating on high-resolution frames, allowing for high efficiency and accuracy at the same time. We demonstrate that our approach achieves a reduction by $3\sim4$ times in FLOPs and $\sim2$ times in memory usage compared to the baseline. This enables training deeper models with more input frames under the same computational budget. To further obviate the need for large-scale 3D convolutions, a temporal aggregation module is proposed to model temporal dependencies in a video at very small additional computational costs. Our models achieve strong performance on several action recognition benchmarks including Kinetics, Something-Something and Moments-in-time. The code and models are available at https://github.com/IBM/bLVNet-TAM.
In this paper, we evaluate training of deep recurrent neural networks with half-precision floats. We implement a distributed, data-parallel, synchronous training algorithm by integrating TensorFlow and CUDA-aware MPI to enable execution across multiple GPU nodes and making use of high-speed interconnects. We introduce a learning rate schedule facilitating neural network convergence at up to $O(100)$ workers. Strong scaling tests performed on clusters of NVIDIA Pascal P100 GPUs show linear runtime and logarithmic communication time scaling for both single and mixed precision training modes. Performance is evaluated on a scientific dataset taken from the Joint European Torus (JET) tokamak, containing multi-modal time series of sensory measurements leading up to deleterious events called plasma disruptions, and the benchmark Large Movie Review Dataset~\cite{imdb}. Half-precision significantly reduces memory and network bandwidth, allowing training of state-of-the-art models with over 70 million trainable parameters while achieving a comparable test set performance as single precision.
Moving Object Detection (MOD) is a critical task for autonomous vehicles as moving objects represent higher collision risk than static ones. The trajectory of the ego-vehicle is planned based on the future states of detected moving objects. It is quite challenging as the ego-motion has to be modelled and compensated to be able to understand the motion of the surrounding objects. In this work, we propose a real-time end-to-end CNN architecture for MOD utilizing spatio-temporal context to improve robustness. We construct a novel time-aware architecture exploiting temporal motion information embedded within sequential images in addition to explicit motion maps using optical flow images.We demonstrate the impact of our algorithm on KITTI dataset where we obtain an improvement of 8% relative to the baselines. We compare our algorithm with state-of-the-art methods and achieve competitive results on KITTI-Motion dataset in terms of accuracy at three times better run-time. The proposed algorithm runs at 23 fps on a standard desktop GPU targeting deployment on embedded platforms.
Recurrent neural networks (RNN) have been successfully applied to various sequential decision-making tasks, natural language processing applications, and time-series predictions. Such networks are usually trained through back-propagation through time (BPTT) which is prohibitively expensive, especially when the length of the time dependencies and the number of hidden neurons increase. To reduce the training time, extreme learning machines (ELMs) have been recently applied to RNN training, reaching a 99\% speedup on some applications. Due to its non-iterative nature, ELM training, when parallelized, has the potential to reach higher speedups than BPTT. In this work, we present \opt, an optimized parallel RNN training algorithm based on ELM that takes advantage of the GPU shared memory and of parallel QR factorization algorithms to efficiently reach optimal solutions. The theoretical analysis of the proposed algorithm is presented on six RNN architectures, including LSTM and GRU, and its performance is empirically tested on ten time-series prediction applications. \opt~is shown to reach up to 845 times speedup over its sequential counterpart and to require up to 20x less time to train than parallel BPTT.
The leading approaches in language modeling are all obsessed with TV shows of my youth - namely Transformers and Sesame Street. Transformers this, Transformers that, and over here a bonfire worth of GPU-TPU-neuromorphic wafer scale silicon. We opt for the lazy path of old and proven techniques with a fancy crypto inspired acronym: the Single Headed Attention RNN (SHA-RNN). The author's lone goal is to show that the entire field might have evolved a different direction if we had instead been obsessed with a slightly different acronym and slightly different result. We take a previously strong language model based only on boring LSTMs and get it to within a stone's throw of a stone's throw of state-of-the-art byte level language model results on enwik8. This work has undergone no intensive hyperparameter optimization and lived entirely on a commodity desktop machine that made the author's small studio apartment far too warm in the midst of a San Franciscan summer. The final results are achievable in plus or minus 24 hours on a single GPU as the author is impatient. The attention mechanism is also readily extended to large contexts with minimal computation. Take that Sesame Street.
Machine Learning as a Service (MLaaS) has become a growing trend in recent years and several such services are currently offered. MLaaS is essentially a set of services that provides machine learning tools and capabilities as part of cloud computing services. In these settings, the cloud has pre-trained models that are deployed and large computing capacity whereas the clients can use these models to make predictions without having to worry about maintaining the models and the service. However, the main concern with MLaaS is the privacy of the client's data. Although there have been several proposed approaches in the literature to run machine learning models on encrypted data, the performance is still far from being satisfactory for practical use. In this paper, we aim to accelerate the performance of running machine learning on encrypted data using combination of Fully Homomorphic Encryption (FHE), Convolutional Neural Networks (CNNs) and Graphics Processing Units (GPUs). We use a number of optimization techniques, and efficient GPU-based implementation to achieve high performance. We evaluate a CNN whose architecture is similar to AlexNet to classify homomorphically encrypted samples from the Cars Overhead With Context (COWC) dataset. To the best of our knowledge, it is the first time such a complex network and large dataset is evaluated on encrypted data. Our approach achieved reasonable classification accuracy of 95% for the COWC dataset. In terms of performance, our results show that we could achieve several thousands times speed up when we implement GPU-accelerated FHE operations on encrypted floating point numbers.
This paper introduces SuperGlue, a neural network that matches two sets of local features by jointly finding correspondences and rejecting non-matchable points. Assignments are estimated by solving a differentiable optimal transport problem, whose costs are predicted by a graph neural network. We introduce a flexible context aggregation mechanism based on attention, enabling SuperGlue to reason about the underlying 3D scene and feature assignments jointly. Compared to traditional, hand-designed heuristics, our technique learns priors over geometric transformations and regularities of the 3D world through end-to-end training from image pairs. SuperGlue outperforms other learned approaches and achieves state-of-the-art results on the task of pose estimation in challenging real-world indoor and outdoor environments. The proposed method performs matching in real-time on a modern GPU and can be readily integrated into modern SfM or SLAM systems. The code and trained weights are publicly available at https://github.com/magicleap/SuperGluePretrainedNetwork.
Self-supervised monocular depth estimation methods generally suffer the occlusion fading issue due to the lack of supervision by the per pixel ground truth. Although a post-processing method was proposed by Godard et. al. to reduce the occlusion fading, the compensated results have a severe halo effect. In this paper, we propose a novel Edge-Guided post-processing to reduce the occlusion fading issue for self-supervised monocular depth estimation. We further introduce Atrous Spatial Pyramid Pooling (ASPP) into the network to reduce the computational costs and improve the inference performance. The proposed ASPP-based network is lighter, faster, and better than current commonly used depth estimation networks. This light-weight network only needs 8.1 million parameters and can achieve up to 40 frames per second for $256\times512$ input in the inference stage using a single nVIDIA GTX1080 GPU. The proposed network also outperforms the current state-of-the-art on the KITTI benchmarks. The ASPP-based network and Edge-Guided post-processing produce better results either quantitatively and qualitatively than the competitors.