Inspecting the information encoded in hidden representations of large language models (LLMs) can explain models' behavior and verify their alignment with human values. Given the capabilities of LLMs in generating human-understandable text, we propose leveraging the model itself to explain its internal representations in natural language. We introduce a framework called Patchscopes and show how it can be used to answer a wide range of questions about an LLM's computation. We show that prior interpretability methods based on projecting representations into the vocabulary space and intervening on the LLM computation can be viewed as instances of this framework. Moreover, several of their shortcomings such as failure in inspecting early layers or lack of expressivity can be mitigated by Patchscopes. Beyond unifying prior inspection techniques, Patchscopes also opens up new possibilities such as using a more capable model to explain the representations of a smaller model, and unlocks new applications such as self-correction in multi-hop reasoning.
Automated segmentation of distinct tumor regions is critical for accurate diagnosis and treatment planning in pediatric brain tumors. This study evaluates the efficacy of the Multi-Planner U-Net (MPUnet) approach in segmenting different tumor subregions across three challenging datasets: Pediatrics Tumor Challenge (PED), Brain Metastasis Challenge (MET), and Sub-Sahara-Africa Adult Glioma (SSA). These datasets represent diverse scenarios and anatomical variations, making them suitable for assessing the robustness and generalization capabilities of the MPUnet model. By utilizing multi-planar information, the MPUnet architecture aims to enhance segmentation accuracy. Our results show varying performance levels across the evaluated challenges, with the tumor core (TC) class demonstrating relatively higher segmentation accuracy. However, variability is observed in the segmentation of other classes, such as the edema and enhancing tumor (ET) regions. These findings emphasize the complexity of brain tumor segmentation and highlight the potential for further refinement of the MPUnet approach and inclusion of MRI more data and preprocessing.
Three-dimensional data have become increasingly present in earth observation over the last decades. However, many 3D surveys are still underexploited due to the lack of accessible and explainable automatic classification methods, for example, new topo-bathymetric lidar data. In this work, we introduce explainable machine learning for 3D data classification using Multiple Attributes, Scales, and Clouds under 3DMASC, a new workflow. This workflow introduces multi-cloud classification through dual-cloud features, encrypting local spectral and geometrical ratios and differences. 3DMASC uses classical multi-scale descriptors adapted to all types of 3D point clouds and new ones based on their spatial variations. In this paper, we present the performances of 3DMASC for multi-class classification of topo-bathymetric lidar data in coastal and fluvial environments. We show how multivariate and embedded feature selection allows the building of optimized predictor sets of reduced complexity, and we identify features particularly relevant for coastal and riverine scene descriptions. Our results show the importance of dual-cloud features, lidar return-based attributes averaged over specific scales, and of statistics of dimensionality-based and spectral features. Additionally, they indicate that small to medium spherical neighbourhood diameters (<7 m) are sufficient to build effective classifiers, namely when combined with distance-to-ground or distance-to-water-surface features. Without using optional RGB information, and with a maximum of 37 descriptors, we obtain classification accuracies between 91 % for complex multi-class tasks and 98 % for lower-level processing using models trained on less than 2000 samples per class. Comparisons with classical point cloud classification methods show that 3DMASC features have a significantly improved descriptive power. Our contributions are made available through a plugin in the CloudCompare software, allowing non-specialist users to create classifiers for any type of 3D data characterized by 1 or 2 point clouds (airborne or terrestrial lidar, structure from motion), and two labelled topo-bathymetric lidar datasets, available on https://opentopography.org/.
Knowledge-based Visual Question Answering about Named Entities is a challenging task that requires retrieving information from a multimodal Knowledge Base. Named entities have diverse visual representations and are therefore difficult to recognize. We argue that cross-modal retrieval may help bridge the semantic gap between an entity and its depictions, and is foremost complementary with mono-modal retrieval. We provide empirical evidence through experiments with a multimodal dual encoder, namely CLIP, on the recent ViQuAE, InfoSeek, and Encyclopedic-VQA datasets. Additionally, we study three different strategies to fine-tune such a model: mono-modal, cross-modal, or joint training. Our method, which combines mono-and cross-modal retrieval, is competitive with billion-parameter models on the three datasets, while being conceptually simpler and computationally cheaper.
Quantum computing networks enable scalable collaboration and secure information exchange among multiple classical and quantum computing nodes while executing large-scale generative AI computation tasks and advanced quantum algorithms. Quantum computing networks overcome limitations such as the number of qubits and coherence time of entangled pairs and offer advantages for generative AI infrastructure, including enhanced noise reduction through distributed processing and improved scalability by connecting multiple quantum devices. However, efficient resource allocation in quantum computing networks is a critical challenge due to factors including qubit variability and network complexity. In this article, we propose an intelligent resource allocation framework for quantum computing networks to improve network scalability with minimized resource costs. To achieve scalability in quantum computing networks, we formulate the resource allocation problem as stochastic programming, accounting for the uncertain fidelities of qubits and entangled pairs. Furthermore, we introduce state-of-the-art reinforcement learning (RL) algorithms, from generative learning to quantum machine learning for optimal quantum resource allocation to resolve the proposed stochastic resource allocation problem efficiently. Finally, we optimize the resource allocation in heterogeneous quantum computing networks supporting quantum generative learning applications and propose a multi-agent RL-based algorithm to learn the optimal resource allocation policies without prior knowledge.
Artificial intelligence (AI) has acquired notorious relevance in modern computing as it effectively solves complex tasks traditionally done by humans. AI provides methods to represent and infer knowledge, efficiently manipulate texts and learn from vast amount of data. These characteristics are applicable in many activities that human find laborious or repetitive, as is the case of the analysis of scientific literature. Manually preparing and writing a systematic literature review (SLR) takes considerable time and effort, since it requires planning a strategy, conducting the literature search and analysis, and reporting the findings. Depending on the area under study, the number of papers retrieved can be of hundreds or thousands, meaning that filtering those relevant ones and extracting the key information becomes a costly and error-prone process. However, some of the involved tasks are repetitive and, therefore, subject to automation by means of AI. In this paper, we present a survey of AI techniques proposed in the last 15 years to help researchers conduct systematic analyses of scientific literature. We describe the tasks currently supported, the types of algorithms applied, and available tools proposed in 34 primary studies. This survey also provides a historical perspective of the evolution of the field and the role that humans can play in an increasingly automated SLR process.
Fusion of a panchromatic (PAN) image and corresponding multispectral (MS) image is also known as pansharpening, which aims to combine abundant spatial details of PAN and spectral information of MS. Due to the absence of high-resolution MS images, available deep-learning-based methods usually follow the paradigm of training at reduced resolution and testing at both reduced and full resolution. When taking original MS and PAN images as inputs, they always obtain sub-optimal results due to the scale variation. In this paper, we propose to explore the self-supervised representation of pansharpening by designing a cross-predictive diffusion model, named CrossDiff. It has two-stage training. In the first stage, we introduce a cross-predictive pretext task to pre-train the UNet structure based on conditional DDPM, while in the second stage, the encoders of the UNets are frozen to directly extract spatial and spectral features from PAN and MS, and only the fusion head is trained to adapt for pansharpening task. Extensive experiments show the effectiveness and superiority of the proposed model compared with state-of-the-art supervised and unsupervised methods. Besides, the cross-sensor experiments also verify the generalization ability of proposed self-supervised representation learners for other satellite's datasets. We will release our code for reproducibility.
There has been a lot of work in question generation where different methods to provide target answers as input, have been employed. This experimentation has been mostly carried out for RNN based models. We use three different methods and their combinations for incorporating answer information and explore their effect on several automatic evaluation metrics. The methods that are used are answer prompting, using a custom product method using answer embeddings and encoder outputs, choosing sentences from the input paragraph that have answer related information, and using a separate cross-attention attention block in the decoder which attends to the answer. We observe that answer prompting without any additional modes obtains the best scores across rouge, meteor scores. Additionally, we use a custom metric to calculate how many of the generated questions have the same answer, as the answer which is used to generate them.
This paper develops a graph reinforcement learning approach to online planning of the schedule and destinations of electric aircraft that comprise an urban air mobility (UAM) fleet operating across multiple vertiports. This fleet scheduling problem is formulated to consider time-varying demand, constraints related to vertiport capacity, aircraft capacity and airspace safety guidelines, uncertainties related to take-off delay, weather-induced route closures, and unanticipated aircraft downtime. Collectively, such a formulation presents greater complexity, and potentially increased realism, than in existing UAM fleet planning implementations. To address these complexities, a new policy architecture is constructed, primary components of which include: graph capsule conv-nets for encoding vertiport and aircraft-fleet states both abstracted as graphs; transformer layers encoding time series information on demand and passenger fare; and a Multi-head Attention-based decoder that uses the encoded information to compute the probability of selecting each available destination for an aircraft. Trained with Proximal Policy Optimization, this policy architecture shows significantly better performance in terms of daily averaged profits on unseen test scenarios involving 8 vertiports and 40 aircraft, when compared to a random baseline and genetic algorithm-derived optimal solutions, while being nearly 1000 times faster in execution than the latter.
With the explosive growth of video data in real-world applications, a comprehensive representation of videos becomes increasingly important. In this paper, we address the problem of video scene recognition, whose goal is to learn a high-level video representation to classify scenes in videos. Due to the diversity and complexity of video contents in realistic scenarios, this task remains a challenge. Most existing works identify scenes for videos only from visual or textual information in a temporal perspective, ignoring the valuable information hidden in single frames, while several earlier studies only recognize scenes for separate images in a non-temporal perspective. We argue that these two perspectives are both meaningful for this task and complementary to each other, meanwhile, externally introduced knowledge can also promote the comprehension of videos. We propose a novel two-stream framework to model video representations from multiple perspectives, i.e. temporal and non-temporal perspectives, and integrate the two perspectives in an end-to-end manner by self-distillation. Besides, we design a knowledge-enhanced feature fusion and label prediction method that contributes to naturally introducing knowledge into the task of video scene recognition. Experiments conducted on a real-world dataset demonstrate the effectiveness of our proposed method.