AI powered code-recommendation systems, such as Copilot and CodeWhisperer, provide code suggestions inside a programmer's environment (e.g., an IDE) with the aim to improve their productivity. Since, in these scenarios, programmers accept and reject suggestions, ideally, such a system should use this feedback in furtherance of this goal. In this work, we leverage prior data of programmers interacting with GitHub Copilot, a system used by millions of programmers, to develop interventions that can save programmer time. We propose a utility theory framework, which models this interaction with programmers and decides which suggestions to display. Our framework Conditional suggestion Display from Human Feedback (CDHF), relies on a cascade of models that predict suggestion acceptance to selectively hide suggestions reducing both latency and programmer verification time. Using data from 535 programmers, we perform a retrospective evaluation of CDHF and show that we can avoid displaying a significant fraction of suggestions that would have been rejected doing so without total knowledge of the suggestions themselves. We further demonstrate the importance of incorporating the programmer's latent unobserved state in deciding when to display suggestions through ablations on user study data. Finally, we showcase that using suggestion acceptance as a reward signal to know which suggestions to display leads to reduced quality suggestions indicating an unexpected pitfall.
相關內容
To ease the programmability of PIM architectures, we propose DaPPA(data-parallel processing-in-memory architecture), a framework that can, for a given application, automatically distribute input and gather output data, handle memory management, and parallelize work across the DPUs. The key idea behind DaPPA is to remove the responsibility of managing hardware resources from the programmer by providing an intuitive data-parallel pattern-based programming interface that abstracts the hardware components of the UPMEM system. Using this key idea, DaPPA transforms a data-parallel pattern-based application code into the appropriate UPMEM-target code, including the required APIs for data management and code partition, which can then be compiled into a UPMEM-based binary transparently from the programmer. While generating UPMEM-target code, DaPPA implements several code optimizations to improve end-to-end performance.
The Self-Sovereign Identity (SSI) is a decentralized paradigm enabling full control over the data used to build and prove the identity. In Internet of Things networks with security requirements, the Self-Sovereign Identity can play a key role and bring benefits with respect to centralized identity solutions. The challenge is to make the SSI compatible with resource-constraint IoT networks. In line with this objective, the paper proposes and discusses an alternative (mutual) authentication process for IoT nodes under the same administration domain. The main idea is to combine the Decentralized IDentifier (DID)-based verification of private key ownership with the verification of a proof that the DID belongs to an evolving trusted set. The solution is built around the proof of membership notion. The paper analyzes two membership solutions, a novel solution designed by the Authors based on Merkle trees and a second one based on the adaptation of Boneh, Boyen and Shacham (BBS) group signature scheme. The paper concludes with a performance estimation and a comparative analysis.
Large Language models (LLMs) possess the capability to engage In-context Learning (ICL) by leveraging a few demonstrations pertaining to a new downstream task as conditions. However, this particular learning paradigm suffers from high instability stemming from substantial variances induced by factors such as the input distribution of selected examples, their ordering, and prompt formats. In this work, we demonstrate that even when all these factors are held constant, the random selection of examples still results in high variance. Consequently, we aim to explore the informative ability of data examples by quantifying the Information Gain (IG) obtained in prediction after observing a given example candidate. Then we propose to sample those with maximum IG. Additionally, we identify the presence of template bias, which can lead to unfair evaluations of IG during the sampling process. To mitigate this bias, we introduce Calibration Before Sampling strategy. The experimental results illustrate that our proposed method can yield an average relative improvement of 14.3% across six classification tasks using three LLMs.
Speech Command Recognition (SCR), which deals with identification of short uttered speech commands, is crucial for various applications, including IoT devices and assistive technology. Despite the promise shown by Convolutional Neural Networks (CNNs) in SCR tasks, their efficacy relies heavily on hyper-parameter selection, which is typically laborious and time-consuming when done manually. This paper introduces a hyper-parameter selection method for CNNs based on the Differential Evolution (DE) algorithm, aiming to enhance performance in SCR tasks. Training and testing with the Google Speech Command (GSC) dataset, the proposed approach showed effectiveness in classifying speech commands. Moreover, a comparative analysis with Genetic Algorithm based selections and other deep CNN (DCNN) models highlighted the efficiency of the proposed DE algorithm in hyper-parameter selection for CNNs in SCR tasks.
Case-based reasoning (CBR) as a methodology for problem-solving can use any appropriate computational technique. This position paper argues that CBR researchers have somewhat overlooked recent developments in deep learning and large language models (LLMs). The underlying technical developments that have enabled the recent breakthroughs in AI have strong synergies with CBR and could be used to provide a persistent memory for LLMs to make progress towards Artificial General Intelligence.
Machine Learning as a Service (MLaaS) APIs provide ready-to-use and high-utility encoders that generate vector representations for given inputs. Since these encoders are very costly to train, they become lucrative targets for model stealing attacks during which an adversary leverages query access to the API to replicate the encoder locally at a fraction of the original training costs. We propose Bucks for Buckets (B4B), the first active defense that prevents stealing while the attack is happening without degrading representation quality for legitimate API users. Our defense relies on the observation that the representations returned to adversaries who try to steal the encoder's functionality cover a significantly larger fraction of the embedding space than representations of legitimate users who utilize the encoder to solve a particular downstream task.vB4B leverages this to adaptively adjust the utility of the returned representations according to a user's coverage of the embedding space. To prevent adaptive adversaries from eluding our defense by simply creating multiple user accounts (sybils), B4B also individually transforms each user's representations. This prevents the adversary from directly aggregating representations over multiple accounts to create their stolen encoder copy. Our active defense opens a new path towards securely sharing and democratizing encoders over public APIs.
In this paper, we propose two new classes of binary array codes, termed V-ETBR and V-ESIP codes, by reformulating and generalizing the variant technique of deriving the well-known generalized row-diagonal parity~(RDP) codes from shortened independent parity~(IP) codes. The V-ETBR and V-ESIP codes are both based on binary parity-check matrices and are essentially variants of two classes of codes over a special polynomial ring (termed ETBR and ESIP codes in this paper). To explore the conditions that make the variant codes binary Maximum Distance Separable~(MDS) array codes that achieve optimal storage efficiency, this paper derives the connections between V-ETBR/V-ESIP codes and ETBR/ESIP codes. These connections are beneficial for constructing various forms of the variant codes. By utilizing these connections, this paper also explicitly presents the constructions of V-ETBR and V-ESIP MDS array codes with any number of parity columns $r$, along with their fast syndrome computations. In terms of construction, all proposed MDS array codes have an exponentially growing total number of data columns with respect to the column size, while alternative codes have that only with linear order. In terms of computation, the proposed syndrome computations make the corresponding encoding/decoding asymptotically require $\lfloor \lg r \rfloor+1$ XOR~(exclusive OR) operations per data bit, when the total number of data columns approaches infinity. This is also the lowest known asymptotic complexity in MDS codes.
Vision-Language Models (VLMs) such as CLIP are trained on large amounts of image-text pairs, resulting in remarkable generalization across several data distributions. The prohibitively expensive training and data collection/curation costs of these models make them valuable Intellectual Property (IP) for organizations. This motivates a vendor-client paradigm, where a vendor trains a large-scale VLM and grants only input-output access to clients on a pay-per-query basis in a black-box setting. The client aims to minimize inference cost by distilling the VLM to a student model using the limited available task-specific data, and further deploying this student model in the downstream application. While naive distillation largely improves the In-Domain (ID) accuracy of the student, it fails to transfer the superior out-of-distribution (OOD) generalization of the VLM teacher using the limited available labeled images. To mitigate this, we propose Vision-Language to Vision-Align, Distill, Predict (VL2V-ADiP), which first aligns the vision and language modalities of the teacher model with the vision modality of a pre-trained student model, and further distills the aligned VLM embeddings to the student. This maximally retains the pre-trained features of the student, while also incorporating the rich representations of the VLM image encoder and the superior generalization of the text embeddings. The proposed approach achieves state-of-the-art results on the standard Domain Generalization benchmarks in a black-box teacher setting, and also when weights of the VLM are accessible.
To protect cryptographic implementations from side-channel vulnerabilities, developers must adopt constant-time programming practices. As these can be error-prone, many side-channel detection tools have been proposed. Despite this, such vulnerabilities are still manually found in cryptographic libraries. While a recent paper by Jancar et al. shows that developers rarely perform side-channel detection, it is unclear if existing detection tools could have found these vulnerabilities in the first place. To answer this question, we surveyed the literature to build a classification of 34 side-channel detection frameworks. The classification we offer compares multiple criteria, including the methods used, the scalability of the analysis or the threat model considered. We then built a unified common benchmark of representative cryptographic operations on a selection of 5 promising detection tools. This benchmark allows us to better compare the capabilities of each tool, and the scalability of their analysis. Additionally, we offer a classification of recently published side-channel vulnerabilities. We then test each of the selected tools on benchmarks reproducing a subset of these vulnerabilities as well as the context in which they appear. We find that existing tools can struggle to find vulnerabilities for a variety of reasons, mainly the lack of support for SIMD instructions, implicit flows, and internal secret generation. Based on our findings, we develop a set of recommendations for the research community and cryptographic library developers, with the goal to improve the effectiveness of side-channel detection tools.
The mobile robot relies on SLAM (Simultaneous Localization and Mapping) to provide autonomous navigation and task execution in complex and unknown environments. However, it is hard to develop a dedicated algorithm for mobile robots due to dynamic and challenging situations, such as poor lighting conditions and motion blur. To tackle this issue, we propose a tightly-coupled LiDAR-visual SLAM based on geometric features, which includes two sub-systems (LiDAR and monocular visual SLAM) and a fusion framework. The fusion framework associates the depth and semantics of the multi-modal geometric features to complement the visual line landmarks and to add direction optimization in Bundle Adjustment (BA). This further constrains visual odometry. On the other hand, the entire line segment detected by the visual subsystem overcomes the limitation of the LiDAR subsystem, which can only perform the local calculation for geometric features. It adjusts the direction of linear feature points and filters out outliers, leading to a higher accurate odometry system. Finally, we employ a module to detect the subsystem's operation, providing the LiDAR subsystem's output as a complementary trajectory to our system while visual subsystem tracking fails. The evaluation results on the public dataset M2DGR, gathered from ground robots across various indoor and outdoor scenarios, show that our system achieves more accurate and robust pose estimation compared to current state-of-the-art multi-modal methods.
小貼士
登錄享
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191