Coming Soon!

Conventional Neural Architecture Search (NAS) aims at finding a single architecture that achieves the best performance, which usually optimizes task related learning objectives such as accuracy. However, a single architecture may not be representative enough for the whole dataset with high diversity and variety. Intuitively, electing domain-expert architectures that are proficient in domain-specific features can further benefit architecture related objectives such as latency. In this paper, we propose InstaNAS—an instanceaware NAS framework—that employs a controller trained to search for a “distribution of architectures” instead of a single architecture; This allows the model to use sophisticated architectures for the difficult samples, which usually comes with large architecture related cost, and shallow architectures for those easy samples. During the inference phase, the controller assigns each of the unseen input samples with a domain expert architecture that can achieve high accuracy with customized inference costs. Experiments within a search space inspired by MobileNetV2 show InstaNAS can achieve up to 48.8% latency reduction without compromising accuracy on a series of datasets against MobileNetV2.

Recent breakthroughs in Neural Architectural Search (NAS) have achieved state-of-the-art performance in many tasks such as image classification and language understanding. However, most existing works only optimize for model accuracy and largely ignore other important factors imposed by the underlying hardware and devices, such as latency and energy, when making inference. In this paper, we first introduce the problem of NAS and provide a survey on recent works. Then we deep dive into two recent advancements on extending NAS into multiple-objective frameworks: MONAS and DPP-Net. Both MONAS and DPP-Net are capable of optimizing accuracy and other objectives imposed by devices, searching for neural architectures that can be best deployed on a wide spectrum of devices: from embedded systems and mobile devices to workstations. Experimental results are poised to show that architectures found by MONAS and DPP-Net achieves Pareto optimality w.r.t the given objectives for various devices.

In this paper, we investigate the use of an unsupervised label clustering technique and demonstrate that it enables substantial improvements in visual relationship prediction accuracy on the Person in Context (PIC) dataset. We propose to group object labels with similar patterns of relationship distribution in the dataset into fewer categories. Label clustering not only mitigates both the large classification space and class imbalance issues, but also potentially increases data samples for each clustered category. We further propose to incorporate depth information as an additional feature into the instance segmentation model. The additional depth prediction path supplements the relationship prediction model in a way that bounding boxes or segmentation masks are unable to deliver. We have rigorously evaluated the proposed techniques and performed various ablation analysis to validate the benefits of them.

Recent breakthroughs in Neural Architectural Search (NAS) have achieved state-of-the-art performances in applications such as image classification and language modeling. However, these techniques typically ignore device-related objectives such as inference time, memory usage, and power consumption. Optimizing neural architecture for device-related objectives is immensely crucial for deploying deep networks on portable devices with limited computing resources. We propose DPP-Net: Device-aware Progressive Search for Pareto-optimal Neural Architectures, optimizing for both device-related (e.g., inference time and memory usage) and device-agnostic (e.g., accuracy and model size) objectives. DPP-Net employs a compact search space inspired by current state-of-the-art mobile CNNs, and further improves search efficiency by adopting progressive search (Liu et al. 2017). Experimental results on CIFAR-10 are poised to demonstrate the effectiveness of Pareto-optimal networks found by DPP-Net, for three different devices: (1) a workstation with Titan X GPU, (2) NVIDIA Jetson TX1 embedded system, and (3) mobile phone with ARM Cortex-A53. Compared to CondenseNet and NASNet (Mobile), DPP-Net achieves better performances: higher accuracy and shorter inference time on various devices. Additional experimental results show that models found by DPP-Net also achieve considerably-good performance on ImageNet as well.

Recent breakthroughs in Neural Architectural Search (NAS) have achieved state-of-the-art performances in many applications such as image recognition. However, these techniques typically ignore platform-related constrictions (e.g., inference time and power consumptions) that can be critical for portable devices with limited computing resources. We propose PPP-Net: a multi-objective architectural search framework to automatically generate networks that achieve Pareto Optimality. PPP-Net employs a compact search space inspired by operations used in state-of-the-art mobile CNNs. PPP-Net has also adopted the progressive search strategy used in a recent literature (Liu et al. (2017a)). Experimental results demonstrate that PPP-Net achieves better performances in both (a) higher accuracy and (b) shorter inference time, comparing to the state-of-the-art CondenseNet.

Modern IoT analytics are computational and data intensive. Existing analytics are mostly hosted in cloud data centers, and may suffer from high latency, network congestion, and privacy issues. In this paper, we design, implement, and evaluate a fog computing platform that runs analytics in a distributed way on multiple devices, including IoT devices, edge servers, and data-center servers. We focus on the core optimization problem: making deployment decisions to maximize the number of satisfied IoT analytics. We carefully formulate the deployment problem and design an efficient algorithm, named SSE, to solve it. Moreover, we conduct a detailed measurement study to derive system models of the IoT analytics based on diverse QoS levels and heterogeneous devices to facilitate the optimal deployment decisions. We implement a testbed to conduct experiments, which show that the system models achieve reasonably good accuracy. More importantly, 100% of the deployed IoT analytics satisfy the QoS targets. We also conduct extensive simulations for largerscale scenarios. The simulation results reveal that our SSE algorithm outperforms a state-of-the-art algorithm by up to 89.4% and 168.3% in terms of the number of satisfied IoT analytics and active devices. In addition, our SSE algorithm reduces CPU, RAM, and network resource consumptions by 18.4%, 12.7%, and 898.3%, respectively, and terminates in polynomial time.</div> </td> </tr>

Modern Internet-of-Things (IoT) applications produce large amount of data and require powerful analytics approaches, such as Deep Learning to extract useful information. Existing IoT applications transmit the data to resource-rich data centers for analytics. However, it may congest networks, overload data centers, and increase security vulnerability. In this paper, we implement a platform, which integrates resources from data centers (servers) to end devices (IoT devices). We launch distributed analytics applications among the devices without sending everything to the data centers. We analyze challenges to implement such a platform and carefully adopt popular open-source projects to overcome the challenges. We then conduct comprehensive experiments on the implemented platform. The results show: (i) the benefits/limitations of distributed analytics, (ii) the importance of decisions on distributing an application across multiple devices, and (iii) the overhead caused by different components in our platform.</div> </td> </tr>