Large Language Models (LLMs) are becoming essential tools for various natural language processing tasks but often suffer from generating outdated or incorrect information. Retrieval-Augmented Generation (RAG) addresses this issue by incorporating external, real-time information retrieval to ground LLM responses. However, the existing RAG systems frequently struggle with the quality of retrieval documents, as irrelevant or noisy documents degrade performance, increase computational overhead, and undermine response reliability. To tackle this problem, we propose Multi-Agent Filtering Retrieval-Augmented Generation (MAIN-RAG), a training-free RAG framework that leverages multiple LLM agents to collaboratively filter and score retrieved documents. Specifically, MAIN-RAG introduces an adaptive filtering mechanism that dynamically adjusts the relevance filtering threshold based on score distributions, effectively minimizing noise while maintaining high recall of relevant documents. The proposed approach leverages inter-agent consensus to ensure robust document selection without requiring additional training data or fine-tuning. Experimental results across four QA benchmarks demonstrate that MAIN-RAG consistently outperforms traditional RAG approaches, achieving a 2-11% improvement in answer accuracy while reducing the number of irrelevant retrieved documents. Quantitative analysis further reveals that our approach achieves superior response consistency and answer accuracy over baseline methods, offering a competitive and practical alternative to training-based solutions.

Spiking Neural Networks (SNNs) have recently become more popular as a biologically plausible substitute for traditional Artificial Neural Networks (ANNs). SNNs are cost-efficient and deployment-friendly because they process input in both spatial and temporal manner using binary spikes. However, we observe that the information capacity in SNNs is affected by the number of timesteps, leading to an accuracyefficiency tradeoff. In this work, we study a fine-grained adjustment of the number of timesteps in SNNs. Specifically, we treat the number of timesteps as a variable conditioned on different input samples to reduce redundant timesteps for certain data.

Analog physical neural networks, which hold promise for improved energy efficiency and speed compared to digital electronic neural networks, are nevertheless typically operated in a relatively high-power regime so that the signal-to-noise ratio (SNR) is large (>10). What happens if an analog system is instead operated in an ultra-low-power regime, in which the behavior of the system becomes highly stochastic and the noise is no longer a small perturbation on the signal? In this paper, we study this question in the setting of optical neural networks operated in the limit where some layers use only a single photon to cause a neuron activation. Neuron activations in this limit are dominated by quantum noise from the fundamentally probabilistic nature of single-photon detection of weak optical signals. We show that it is possible to train stochastic optical neural networks to perform deterministic image-classification tasks with high accuracy in spite of the extremely high noise (SNR ~ 1) by using a training procedure that directly models the stochastic behavior of photodetection. We experimentally demonstrated MNIST classification with a test accuracy of 98% using an optical neural network with a hidden layer operating in the single-photon regime; the optical energy used to perform the classification corresponds to 0.008 photons per multiply-accumulate (MAC) operation, which is equivalent to 0.003 attojoules of optical energy per MAC. Our experiment used >40x fewer photons per inference than previous state-of-the-art low-optical-energy demonstrations, to achieve the same accuracy of >90%. Our work shows that some extremely stochastic analog systems, including those operating in the limit where quantum noise dominates, can nevertheless be used as layers in neural networks that deterministically perform classification tasks with high accuracy if they are appropriately trained.

Meteorologist Mike Slifer said to expect another chilly start on Tuesday. Then, he tracked possible rain for Thursday. Five children, all from Derby, CT, were killed in a car crash in Scarsdale, NY on Sunday morning. Meteorologist Mike Slifer said the next couple of days would be sunny. Clouds start to thicken on Wednesday.

Leesburg, VA, August 17, 2022--According to ARRS' American Journal of Roentgenology (AJR), machine learning models applied to presently underutilized imaging features could help construct more reliable criteria for organ allocation and liver transplant eligibility. "The findings suggest that machine learning-based models can predict recurrence before therapy allocation in patients with early-stage hepatocellular carcinoma (HCC) initially eligible for liver transplant," wrote corresponding author Julius Chapiro from the department of radiology and biomedical imaging at Yale University School of Medicine in New Haven, CT. Chapiro and colleagues' proof-of-concept study included 120 patients (88 men, 32 women; median age, 60 years) diagnosed with early-stage HCC between June 2005 and March 2018, who were initially eligible for liver transplant and underwent treatment by transplant, resection, or thermal ablation. Patients underwent pretreatment MRI and posttreatment imaging surveillance, and imaging features were extracted from postcontrast phases of pretreatment MRI examinations using a pretrained convolutional neural network (VGG-16). Pretreatment clinical characteristics (including laboratory data) and extracted imaging features were integrated to develop three ML models--clinical, imaging, combined--for recurrence prediction within 1–6 years posttreatment. Ultimately, all three models predicted posttreatment recurrence for early-stage HCC from pretreatment clinical (AUC 0.60–0.78,

According to ARRS' American Journal of Roentgenology (AJR), machine learning models applied to presently underutilized imaging features could help construct more reliable criteria for organ allocation and liver transplant eligibility. "The findings suggest that machine learning-based models can predict recurrence before therapy allocation in patients with early-stage hepatocellular carcinoma (HCC) initially eligible for liver transplant," wrote corresponding author Julius Chapiro from the department of radiology and biomedical imaging at Yale University School of Medicine in New Haven, CT. Chapiro and colleagues' proof-of-concept study included 120 patients (88 men, 32 women; median age, 60 years) diagnosed with early-stage HCC between June 2005 and March 2018, who were initially eligible for liver transplant and underwent treatment by transplant, resection, or thermal ablation. Patients underwent pretreatment MRI and posttreatment imaging surveillance, and imaging features were extracted from postcontrast phases of pretreatment MRI examinations using a pretrained convolutional neural network (VGG-16). Pretreatment clinical characteristics (including laboratory data) and extracted imaging features were integrated to develop three ML models--clinical, imaging, combined--for recurrence prediction within 1–6 years posttreatment. Ultimately, all three models predicted posttreatment recurrence for early-stage HCC from pretreatment clinical (AUC 0.60–0.78,

We outline emerging opportunities and challenges to enhance the utility of AI for scientific discovery. The distinct goals of AI for industry versus the goals of AI for science create tension between identifying patterns in data versus discovering patterns in the world from data. If we address the fundamental challenges associated with "bridging the gap" between domain-driven scientific models and data-driven AI learning machines, then we expect that these AI models can transform hypothesis generation, scientific discovery, and the scientific process itself.

Lawrence currently volunteers as the COVID data scientist on Ridgefield's COVID-19 Task Force, providing daily analysis of the latest COVID-19 data to help town officials make science-based policy decisions, and provides periodic analysis of vaccination rates to the Office of the Governor of Connecticut. Lawrence's work has evolved from nuclear science to computer science to machine learning and, most recently, to quantitative finance. He joined IBM Research in Yorktown Heights, New York, in 1987, where he held a number of management positions, most recently as Distinguished Research Staff Member and Senior Manager, Machine Learning & Decision Analytics. From 2016 to 2019, he was president of PCIX, Inc., a New York City venture capital-funded startup that used machine learning to extract quantitative insight on the relationship between private-equity transactions and the performance of public markets. Lawrence received a Bachelor of Science in Chemical Engineering from Stanford University and a doctorate in Nuclear Engineering from the University of Illinois.

Chapter 5 of this free 15 chapter AI handbook provides an overview of natural language processing.

Given data, finding a faithful low-dimensional hyperbolic embedding of the data is a key method by which we can extract hierarchical information or learn representative geometric features of the data. In this paper, we explore a new method for learning hyperbolic representations by taking a metric-first approach. Rather than determining the low-dimensional hyperbolic embedding directly, we learn a tree structure on the data. This tree structure can then be used directly to extract hierarchical information, embedded into a hyperbolic manifold using Sarkar's construction \cite{sarkar}, or used as a tree approximation of the original metric. To this end, we present a novel fast algorithm \textsc{TreeRep} such that, given a $\delta$-hyperbolic metric (for any $\delta \geq 0$), the algorithm learns a tree structure that approximates the original metric. In the case when $\delta = 0$, we show analytically that \textsc{TreeRep} exactly recovers the original tree structure. We show empirically that \textsc{TreeRep} is not only many orders of magnitude faster than previously known algorithms, but also produces metrics with lower average distortion and higher mean average precision than most previous algorithms for learning hyperbolic embeddings, extracting hierarchical information, and approximating metrics via tree metrics.