NEC Laboratories America

Archive for category: Publications

You are here: / / Publications

Dynamic Prompting: A Unified Framework for Prompt Tuning

/in /by

It has been demonstrated that prompt tuning is highly effective in efficiently eliciting knowledge from language models (LMs). However, the prompt tuning still lags behind fine tuning, especially when the LMs are small. P tuning v2 (Liu et al., 2021b) makes it comparable with finetuning by adding continuous prompts for every layer of the pre trained model. However, prepending fixed soft prompts for all instances, regardless of their discrepancy, is doubtful. In particular, the inserted prompt position, length, and the representations ofprompts for diversified instances through different tasks could all affect the prompt tuning performance. To fill this gap, we propose dynamic prompting (DP): the position, length, and prompt representation can all be dynamically optimized with respect to different tasks and instances. We conduct comprehensive experiments on the SuperGlue benchmark tovalidate our hypothesis and demonstrate substantial improvements. We also derive a unified framework for supporting our dynamic prompting strategy. In particular, we use a simple learning network and Gumble Softmax for learning instance dependent guidance. Experimental results show that simple instance level position aware soft prompts can improve the classification accuracy of up to 6 points on average on five datasets, reducing its gap with fine tuning. Besides, we also prove its universal usefulness under full data, few shot, andmultitask regimes. Combining them together can even further unleash the power of DP, narrowing the distance between fine tuning.

Read more

Field Trial of Coexistence and Simultaneous Switching of Real-time Fiber Sensing and 400GbE Supporting DCI and 5G Mobile Services

/in /by

Coexistence of real-time constant-amplitude distributed acoustic sensing (DAS) and 400GbE signals is verified by field trial over metro fibers, demonstrating no QoT impact during co-propagation and supporting preemptive DAS-informed optical path switching before link failure

Read more

Polarization Sensing Using Polarization Rotation Matrix Eigenvalue Method

/in /by

Polarization-based, multi-span sensing over a link with reflection-back circuits is demonstrated experimentally. By measuring rotation matrices instead of just monitoring polarization, a 35 dB extinction in localization is achieved regardless of the disturbance magnitude.

Read more

Content-aware auto-scaling of stream processing applications on container orchestration platforms

/in /by

Modern applications are designed as an interacting set of microservices, and these applications are typically deployed on container orchestration platforms like Kubernetes. Several attractive features in Kubernetes make it a popular choice for deploying applications, and automatic scaling is one such feature. The default horizontal scaling technique in Kubernetes is the Horizontal Pod Autoscaler (HPA). It scales each microservice independently while ignoring the interactions among the microservices in an application. In this paper, we show that ignoring such interactions by HPA leads to inefficient scaling, and the optimal scaling of different microservices in the application varies as the stream content changes. To automatically adapt to variations in stream content, we present a novel system called DataX AutoScaler that leverages knowledge of the entire stream processing application pipeline to efficiently auto-scale different microservices by taking into account their complex interactions. Through experiments on real-world video analytics applications, such as face recognition and pose classification, we show that DataX AutoScaler adapts to variations in stream content and achieves up to 43% improvement in overall application performance compared to a baseline system that uses HPA.

Read more

Exploring the limits of ChatGPT for Query or Aspect based Text Summarization

/in /by

Text summarization has been a crucial problem in natural language processing (NLP) for several decades. It aims to condense lengthy documents into shorter versions while retaining the most critical information. Various methods have been proposed for text summarization, including extractive and abstractive summarization. The emergence of large language models (LLMs) like GPT3 and ChatGPT has recently created significant interest in using these models for text summarization tasks. Recent studies (Goyal et al., 2022, Zhang et al., 2023) have shown that LLMs generated news summaries are already on par with humans. However, the performance of LLMs for more practical applications like aspect or query based summaries is underexplored. To fill this gap, we conducted an evaluation of ChatGPT’s performance on four widely used benchmark datasets, encompassing diverse summaries from Reddit posts, news articles, dialogue meetings, and stories. Our experiments reveal that ChatGPT’s performance is comparable to traditional fine tuning methods in terms of Rouge scores. Moreover, we highlight some unique differences between ChatGPT generated summaries and human references, providing valuable insights into the superpower of ChatGPT for diverse text summarization tasks. Our findings call for new directions in this area, and we plan to conduct further research to systematically examine the characteristics of ChatGPT generated summaries through extensive human evaluation.

Read more

DAS over 1,007-km Hybrid Link with 10-Tb/s DP-16QAM Co-propagation using Frequency-Diverse Chirped Pulses

/in /by

We report the first distributed acoustic sensing (DAS) experiment with over >1,000 km reach on a hybrid link comprising of a mixture of field and lab fibers with bi-directional inline Raman amplification after each span. We used 20× frequency-diversity chirped-pulses for the probe signal,and recovered the Rayleigh backscatter using a coherent receiver with correlation detection and diversity combining. A measurand resolution of ∼100 pϵ/√ Hz at a gauge length of 20 meters achieved in the offline experiment. We also demonstrate the first real-time FPGA implementation of chirped-pulse DAS without frequency diversity over a range of 210 km.

Read more

Time Series Contrastive Learning with Information-Aware Augmentations

/in /by

Various contrastive learning approaches have been proposed in recent years and have achieved significant empirical success. While effective and prevalent, contrastive learning has been less explored for time series data. A key component of contrastive learning is to select appropriate augmentations, imposing some priors to construct feasible positive samples, such that an encoder can be trained to learn robust and discriminative representations. Unlike image and language domains where “desired” augmented samples can be generated with the rule of thumb guided by prefabricated human priors, the ad-hoc manual selection of time series augmentations is hindered by their diverse and human-unrecognizable temporal structures. How to find the desired augmentations of time series data that are meaningful for given contrastive learning tasks and datasets remains an open question. In this work, we address the problem by encouraging both high fidelity and variety based on information theory. A theoretical analysis leads to the criteria for selecting feasible data augmentations. On top of that, we propose a new contrastive learning approach with information-aware augmentations, InfoTS, that adaptively selects optimal augmentations for time series representation learning. Experiments on various datasets show highly competitive performance with up to a 12.0% reduction in MSE on forecasting tasks and up to 3.7% relative improvement in accuracy on classification tasks over the leading baselines.

Read more

Adversarial Alignment for Source Free Object Detection

/in /by

Source-free object detection (SFOD) aims to transfer a detector pre-trained on a label-rich source domain to an unlabeled target domain without seeing source data. While most existing SFOD methods generate pseudo labels via a source-pretrained model to guide training, these pseudo labels usually contain high noises due to heavy domain discrepancy. In order to obtain better pseudo supervisions, we divide the target domain into source-similar and source-dissimilar parts and align them in the feature space by adversarial learning. Specifically, we design a detection variance-based criterion to divide the target domain. This criterion is motivated by a finding that larger detection variances denote higher recall and larger similarity to the source domain. Then we incorporate an adversarial module into a mean teacher framework to drive the feature spaces of these two subsets indistinguishable. Extensive experiments on multiple cross-domain object detection datasets demonstrate that our proposed method consistently outperforms the compared SFOD methods. Our implementation is available at https://github.com/ChuQiaosong

Read more

Ambient Noise based Weakly Supervised Manhole Localization Methods over Deployed Fiber Networks

/in /by

We present a manhole localization method based on distributed fiber optic sensing and weakly supervised machine learning techniques. For the first time to our knowledge, ambient environment data is used for underground cable mapping with the promise of enhancing operational efficiency and reducing field work. To effectively accommodate the weak informativeness of ambient data, a selective data sampling scheme and an attention-based deep multiple instance classification model are adopted, which only requires weakly annotated data. The proposed approach is validated on field data collected by a fiber sensing system over multiple existing fiber networks.

Read more

Drone Detection and Localization using Enhanced Fiber-Optic Acoustic Sensor and Distributed Acoustic Sensing Technology

/in /by

In recent years, the widespread use of drones has led to serious concerns about safety and privacy. Drone detection using microphone arrays has proven to be a promising method. However, it is challenging for microphones to serve large-scale applications due to the issues of synchronization, complexity, and data management. Moreover, distributed acoustic sensing (DAS) using optical fibers has demonstrated its advantages in monitoring vibrations over long distances but does not have the necessary sensitivity for weak airborne acoustics. In this work, we present, to the best of our knowledge, the first fiber-optic quasi-distributed acoustic sensing demonstration for drone surveillance. We develop enhanced fiber-optic acoustic sensors (FOASs) for DAS to detect drone sound. The FOAS shows an ultra-high measured sensitivity of −101.21 re. 1rad/µPa, as well as the capability for high-fidelity speech recovery. A single DAS can interrogate a series of FOASs over a long distance via optical fiber, enabling intrinsic synchronization and centralized signal processing.We demonstrate the field test of drone detection and localization by concatenating four FOASs as DAS. Both the waveforms and spectral features of the drone sound are recognized. With acoustic field mapping and data fusion, accurate drone localization is achieved with a root-mean-square error (RMSE) of 1.47 degrees. This approach holds great potential in large-scale sound detection applications, such as drone detection or city event monitoring.

Read more