MELON:IndirectPromptInjectionDefenseviaMaskedRe-executionandToolComparison
Recent research has explored that LLM agents are vulnerable to indirect prompt injection (IPI) attac
首页 > 研究论文
-
-
JointMoEScalingLaws:MixtureofExpertsCanBeMemoryEfficient
Mixture of Experts (MoE) architectures have significantly increased computational efficiency in both
-
TransformingSciencewithLargeLanguageModels:ASurveyonAI-assistedScientificDiscovery,Experimentation,ContentGeneration,andEvaluation
With the advent of large multimodal language models, science is now at a threshold of an AI-based te
-
AnAnnotatedReadingof'TheSingerofTales'intheLLMEra
The Parry-Lord oral-formulaic theory was a breakthrough in understanding how oral narrative poetry i
-
LP-DETR:Layer-wiseProgressiveRelationsforObjectDetection
This paper presents LP-DETR (Layer-wise Progressive DETR), a novel approach that enhances DETR-based
-
LatentSwapJointDiffusionforLong-FormAudioGeneration
Previous work on long-form audio generation using global-view diffusion or iterative generation dema
-
AdoptionofAI-AssistedE-Scooters:TheRoleofPerceivedTrust,Safety,andDemographicDrivers
E-scooters have become a more dominant mode of transport in recent years. However, the rise in their
-
"ItFeltLikeIWasLeftintheDark":ExploringInformationNeedsandDesignOpportunitiesforFamilyCaregiversofOlderAdultPatientsinCriticalCareSettings
Older adult patients constitute a rapidly growing subgroup of Intensive Care Unit (ICU) patients. In
-
FlexibleandEfficientGrammar-ConstrainedDecoding
Large Language Models (LLMs) are often asked to generate structured outputs that obey precise syntac
-
ApplE:AnAppliedEthicsOntologywithEventContext
Applied ethics is ubiquitous in most domains, requiring much deliberation due to its philosophical n
-
LeveragingHypernetworksandLearnableKernelsforConsumerEnergyForecastingAcrossDiverseConsumerTypes
Consumer energy forecasting is essential for managing energy consumption and planning, directly infl
-
LearningTemporalInvarianceinAndroidMalwareDetectors
Learning-based Android malware detectors degrade over time due to natural distribution drift caused
-
LostinTime:ClockandCalendarUnderstandingChallengesinMultimodalLLMs
Understanding time from visual representations is a fundamental cognitive skill, yet it remains a ch
-
MitigatingUnintendedMemorizationwithLoRAinFederatedLearningforLLMs
Federated learning (FL) is a popular paradigm for collaborative training which avoids direct data ex
-
Causalitycansystematicallyaddressthemonstersunderthebench(marks)
Effective and reliable evaluation is essential for advancing empirical machine learning. However, th
-
ChallengeMe:AnAdversarialLearning-enabledTextSummarizationFramework
The astonishing performance of large language models (LLMs) and their remarkable achievements in pro
-
AdaptiveGraphofThoughts:Test-TimeAdaptiveReasoningUnifyingChain,Tree,andGraphStructures
Large Language Models (LLMs) have demonstrated impressive reasoning capabilities, yet their performa
-
ComputingandLearningonCombinatorialData
The twenty-first century is a data-driven era where human activities and behavior, physical phenomen
-
Preference-awarecompensationpoliciesforcrowdsourcedon-demandservices
Crowdsourced on-demand services offer benefits such as reduced costs, faster service fulfillment tim
-
DifferentiableMobileDisplayPhotometricStereo
Display photometric stereo uses a display as a programmable light source to illuminate a scene with
