Minimizing the energy consumption of computing is essential to improve efficiency and enable further performance growth. A central point where to address this is in the scheduling of tasks onto multiprocessor systems. In many previous studies the focus of static scheduling algorithms has been on the energy consumption of the processors, where techniques like Dynamic Voltage and Frequency Scaling (DVFS) have been successfully utilized. Despite the large body of work, a notable gap exists: the energy consumed by data transfer, in particular by inter-processor communication, is neglected, even though it becomes more and more significant. To address this, we introduce a revised model that incorporates the energy costs related to task communication, combining it with the existing DVFS techniques and different static energy models. Using the proposed novel energy model, we adjust existing and propose new algorithms and study their performance as influenced by the newly introduced energy parameters. The algorithms cover both major categories, namely : list-scheduling and cluster-scheduling. Using a large set of task graphs of many different structures and sizes, and different parallel systems, our evaluation offers an extensive comparison of the algorithms and provides insights into their ability to conserve energy within the new model’s framework.

Alzheimer's Disease (AD) is a progressive neurodegenerative disorder that leads to dementia, and early intervention can greatly benefit from analyzing linguistic abnormalities.In this work, we explore the potential of Large Language Models (LLMs) as health assistants for AD diagnosis from patient-generated text using in-context learning (ICL), where tasks are defined through a few input-output examples.Empirical results reveal that conventional ICL methods, such as similarity-based selection, perform poorly for AD diagnosis, likely due to the inherent complexity of this task.To address this, we introduce Delta-KNN, a novel demonstration selection strategy that enhances ICL performance.Our method leverages a delta score to assess the relative gains of each training example, coupled with a KNN-based retriever that dynamically selects optimal "representatives" for a given input.Experiments on two AD detection datasets across three opensource LLMs demonstrate that Delta-KNN consistently outperforms existing ICL baselines.Notably, when using the Llama-3.1 model, our approach achieves new state-of-the-art results, surpassing even supervised classifiers. 1

While in-context Learning (ICL) has proven to be an effective technique to improve the performance of Large Language Models (LLMs) in a variety of complex tasks, notably in translating natural language questions into Structured Query Language (NL2SQL), the question of how to select the most beneficial demonstration examples remains an open research problem. While prior works often adapted off-the-shelf encoders to retrieve examples dynamically, an inherent discrepancy exists in the representational capacities between the external retrievers and the LLMs. Further, optimizing the selection of examples is a non-trivial task, since there are no straightforward methods to assess the relative benefits of examples without performing pairwise inference. To address these shortcomings, we propose Detriever, a novel demonstration retrieval framework that learns a weighted combination of LLM hidden states, where rich semantic information is encoded. To train the model, we propose a proxy score that estimates the relative benefits of examples based on the similarities between output queries. Experiments on two popular NL2SQL benchmarks demonstrate that our method significantly outperforms the state-of-the-art baselines for the NL2SQL tasks.

We introduce CEMTM, a context-enhanced multimodal topic model designed to infer coherent and interpretable topic structures from both short and long documents containing text and images.CEMTM builds on fine-tuned large vision language models (LVLMs) to obtain contextualized embeddings, and employs a distributional attention mechanism to weight token-level contributions to topic inference.A reconstruction objective aligns topic-based representations with the document embedding, encouraging semantic consistency across modalities.Unlike existing approaches, CEMTM can process multiple images per document without repeated encoding and maintains interpretability through explicit word-topic and documenttopic distributions.Extensive experiments on six multimodal benchmarks show that CEMTM consistently outperforms unimodal and multimodal baselines, achieving a remarkable average LLM score of 2.61 (1-3 scale).Further analysis shows its effectiveness in downstream few-shot retrieval and its ability to capture visually grounded semantics in complex domains such as scientific articles 1 .

Model Free Reinforcement Learning (MFRL) has shown significant promise for learning dexterous robotic manipulation tasks, at least in simulation. However, the high number of samples, as well as the long training times, prevent MFRL from scaling to complex real-world tasks. Model- Based Reinforcement Learning (MBRL) emerges as a potential solution that, in theory, can improve the data efficiency of MFRL approaches. This could drastically reduce the training time of MFRL, and increase the application of RL for real- world robotic tasks. This article presents a study on the feasibility of using the state-of-the-art MBRL to improve the training time for two real-world dexterous manipulation tasks. The evaluation is conducted on a real low-cost robot gripper where the predictive model and the control policy are learned from scratch. The results indicate that MBRL is capable of learning accurate models of the world, but does not show clear improvements in learning the control policy in the real world as prior literature suggests should be expected.

The multi-head self-attention mechanism of the transformer model has been thoroughly investigated recently.In one vein of study, researchers are interested in understanding why and how transformers work.In another vein, researchers propose new attention augmentation methods to make transformers more accurate, efficient and interpretable.In this paper, we combine these two lines of research in a human-in-the-loop pipeline to first discover important task-specific attention patterns.Then those patterns are injected, not only to smaller models, but also to the original model.The benefits of our pipeline and discovered patterns are demonstrated in two case studies with extractive summarization and topic segmentation.After discovering interpretable patterns in BERT-based models fine-tuned for the two downstream tasks, experiments indicate that when we inject the patterns into attention heads, the models show considerable improvements in accuracy and efficiency.

Torsten Scholak, Raymond Li, Dzmitry Bahdanau, Harm de Vries, Chris Pal. Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2021.

We present a method to produce abstractive summaries of long documents that exceed several thousand words via neural abstractive summarization.We perform a simple extractive step before generating a summary, which is then used to condition the transformer language model on relevant information before being tasked with generating a summary.We also show that this approach produces more abstractive summaries compared to prior work that employs a copy mechanism while still achieving higher ROUGE scores.We provide extensive comparisons with strong baseline methods, prior state of the art work as well as multiple variants of our approach including those using only transformers, only extractive techniques and combinations of the two.We examine these models using four different summarization tasks and datasets: arXiv papers, PubMed papers, the Newsroom and BigPatent datasets.We find that transformer based methods produce summaries with fewer n-gram copies, leading to n-gram copying statistics that are more similar to human generated abstracts.We include a human evaluation, finding that transformers are ranked highly for coherence and fluency, but purely extractive methods score higher for informativeness and relevance.We hope that these architectures and experiments may serve as strong points of comparison for future work.

Recent advances in video processing utilizing deep learning primitives achieved breakthroughs in fundamental problems in video analysis such as frame classification and object detection enabling an array of new applications. In this paper we study the problem of interactive declarative query processing on video streams. In particular we introduce a set of approximate filters to speed up queries that involve objects of specific type (e.g., cars, trucks, etc.) on video frames with associated spatial relationships among them (e.g., car left of truck). The resulting filters are able to assess quickly if the query predicates are true to proceed with further analysis of the frame or otherwise not consider the frame further avoiding costly object detection operations. We propose two classes of filters IC and OD, that adapt principles from deep image classification and object detection. The filters utilize extensible deep neural architectures and are easy to deploy and utilize. In addition, we propose statistical query processing techniques to process aggregate queries involving objects with spatial constraints on video streams and demonstrate experimentally the resulting increased accuracy on the resulting aggregate estimation. Combined these techniques constitute a robust set of video monitoring query processing techniques. We demonstrate that the application of the techniques proposed in conjunction with declarative queries on video streams can dramatically increase the frame processing rate and speed up query processing by at least two orders of magnitude. We present the results of a thorough experimental study utilizing benchmark video data sets at scale demonstrating the performance benefits and the practical relevance of our proposals.

There has been growing interest in using neural networks and deep learning techniques to create dialogue systems. Conversational recommendation is an interesting setting for the scientific exploration of dialogue with natural language as the associated discourse involves goal-driven dialogue that often transforms naturally into more free-form chat. This paper provides two contributions. First, until now there has been no publicly available large-scale data set consisting of real-world dialogues centered around recommendations. To address this issue and to facilitate our exploration here, we have collected ReDial, a data set consisting of over 10,000 conversations centered around the theme of providing movie recommendations. We make this data available to the community for further research. Second, we use this dataset to explore multiple facets of conversational recommendations. In particular we explore new neural architectures, mechanisms and methods suitable for composing conversational recommendation systems. Our dataset allows us to systematically probe model sub-components addressing different parts of the overall problem domain ranging from: sentiment analysis and cold-start recommendation generation to detailed aspects of how natural language is used in this setting in the real world. We combine such sub-components into a full-blown dialogue system and examine its behavior.