Search

Contextual memory integration remains a high challenge in the development of language models, particularly in tasks that require maintaining coherence over extended sequences. Traditional approaches, such as self-attention mechanisms and memory-augmented architectures, often prioritize short-term dependencies, leading to fragmentation and inconsistency in long-range contextual understanding. Inspired by principles of synaptic plasticity observed in biological neural systems, a novel mechanism, Synaptic Resonance, is introduced to dynamically reinforce relevant memory pathways during training and inference. Unlike static memory representations, this mechanism continuously adjusts synaptic weight matrices based on contextual relevance, allowing for improved information retention without excessive computational overhead. Evaluations conducted on an open-source language model demonstrate reductions in perplexity, enhancements in contextual coherence, and increased robustness against input noise, highlighting the effectiveness of reinforcement-driven memory modulation. Comparative analysis against baseline models further reveals that the proposed approach achieves higher memory retention efficiency while maintaining computational feasibility. The architectural modifications integrate seamlessly into existing transformer-based frameworks, ensuring stable convergence and efficient inference without sacrificing scalability. Applications benefiting from improved long-term contextual consistency, such as dialogue systems and document summarization, stand to gain from this approach. Empirical findings suggest that dynamically reinforced memory pathways offer a promising alternative to conventional memory mechanisms, addressing longstanding limitations in extended sequence modeling.

Token prediction stability remains a challenge in autoregressive generative models, where minor variations in early inference steps often lead to significant semantic drift over extended sequences. A structured modulation mechanism was introduced to regulate hidden state transitions, ensuring that latent representation trajectories remain aligned with prior contextual dependencies while preserving generative flexibility. The modulation framework was designed to function within transformer-based architectures, dynamically constraining representation evolution without imposing external memory dependencies or extensive architectural modifications. Empirical evaluations demonstrated that structured latent adjustments contributed to reductions in perplexity fluctuations, entropy variance, and lexical instability, improving coherence in long-form text generation. Gradient propagation stability was further analyzed, revealing that the modulation process led to smoother optimization pathways, mitigating erratic fluctuations in weight updates across successive inference steps. The computational efficiency of the modulation process was assessed, showing that its integration within transformer-based architectures introduced only marginal overhead while maintaining compatibility with existing optimization frameworks. The structured modulation constraints also influenced syntactic variation, preventing excessive repetition while maintaining balanced sentence length distributions. Comparative evaluations against baseline models reinforced the role of controlled latent state evolution in improving pronoun resolution, logical consistency, and contextual alignment across autoregressive text generation tasks.

Why do biological and artificial neurons sometimes modularise, each encoding a single meaningful variable, and sometimes entangle their representation of many variables? In this work, we develop a theory of when biologically inspired networks -- those that are nonnegative and energy efficient -- modularise their representation of source variables (sources). We derive necessary and sufficient conditions on a sample of sources that determine whether the neurons in an optimal biologically-inspired linear autoencoder modularise. Our theory applies to any dataset, extending far beyond the case of statistical independence studied in previous work. Rather we show that sources modularise if their support is ``sufficiently spread''. From this theory, we extract and validate predictions in a variety of empirical studies on how data distribution affects modularisation in nonlinear feedforward and recurrent neural networks trained on supervised and unsupervised tasks. Furthermore, we apply these ideas to neuroscience data, showing that range independence can be used to understand the mixing or modularising of spatial and reward information in entorhinal recordings in seemingly conflicting experiments. Further, we use these results to suggest alternate origins of mixed-selectivity, beyond the predominant theory of flexible nonlinear classification. In sum, our theory prescribes precise conditions on when neural activities modularise, providing tools for inducing and elucidating modular representations in brains and machines., Comment: 40 pages, 16 figures. WD and KH contributed equally; LH and JHL contributed equally

The original proposal of quantum key distribution (QKD) was based on ideal single photon sources, which 40 years later, are still challenging to develop. Therefore, the development of decoy state protocols using weak coherent states (WCS) from lasers, set the frontier in terms of secure key rates. Here, we propose and experimentally emulate two simple-to-implement protocols that allow practical, far from ideal sub-Poissonian photon sources to outperform state-of-the-art WCS. By engineering the photon statistics of a quantum dot's biexciton-exciton cascade, we show that either a truncated decoy state protocol or a heralded purification protocol can be employed to achieve a significantly increased performance in terms of the maximal allowed channel loss for secure key creation, which can exceed that of WCS by more than 3dB. We then show that our recently demonstrated room temperature single photon sources, based on giant colloidal quantum dots coupled to nano-antennas, are already well within the optimal performance range. These protocols can be utilized efficiently on a host of various sub-Poissonian quantum emitters having controllable photon statistics, offering a practical approach to QKD without the hindering requirements on the single photon purity of the photon source.

High-dimensional quantum key distribution (HDQKD) is a promising avenue to address the inherent limitations of basic QKD protocols. However, experimental realizations of HDQKD to date have relied on indeterministic photon sources that limit the achievable key rate. In this paper, we demonstrate a full emulation of a HDQKD system using a single colloidal giant quantum dot (gQD) as a deterministic, compact and room-temperature single-photon source (SPS). We demonstrate a practical protocol by encoding information in a high-dimensional space ($d = 3$) of the orbital angular momentum of the photons. Our experimental configuration incorporates two spatial light modulators for encoding and decoding the spatial information carried by individual photons. Our experimental demonstration establishes the feasibility of utilizing high radiative quantum yield gQDs as practical SPSs for HDQKD. We also demonstrate experimentally secure qudit transmission exceeding one secure bit per photon, thus already beating the traditional d=2 QKD capacity.

Here we report the results of microgravity experiments performed on the International Space Station (ISS) to study crystallized, metastable colloidal liquids in the region of the hard sphere phase diagram that has been found to be glassy on Earth. Using confocal microscopy, we observed the self-assembly of high density, three-dimensional colloidal crystals from micron-size hard spheres suspended within a fluid medium. The largest face-centered cubic (FCC) phase measured 27 x 1.5 x 0.15 mm. From a practical aspect, the fact that a single, topological defect-free FCC colloidal crystal can be grown in microgravity and the crystal returned from orbit suggests new routes for manufacturing colloidal devices, particularly optical elements in space., Comment: 16 pages, 6 figures

To harness the potential of radially polarized single photons in applications such as high-dimensional quantum key distribution (HD-QKD) and quantum communication, we demonstrate an on-chip, room-temperature device, which generates highly directional radially polarized photons at very high rates. The photons are emitted from a giant CdSe/CdS colloidal quantum dot (gQD) accurately positioned at the tip of a metal nanocone centered inside a hybrid metal-dielectric bullseye antenna. We show that due to the large and selective Purcell enhancement specifically for the out-of-plane optical dipole of the gQD, the emitted photons can have a very high degree of radial polarization (>93%), based on a quantitative metric. Our study emphasizes the importance of accurate gQD positioning for optimal radial polarization purity through extensive experiments and simulations, which contribute to the fundamental understanding of radial polarization in nanostructured devices and pave the way for implementation of such systems in practical applications using structured quantum light.

Machine learning (ML) offers a promising solution to pathloss prediction. However, its effectiveness can be degraded by the limited availability of data. To alleviate these challenges, this paper introduces a novel simulation-enhanced data augmentation method for ML pathloss prediction. Our method integrates synthetic data generated from a cellular coverage simulator and independently collected real-world datasets. These datasets were collected through an extensive measurement campaign in different environments, including farms, hilly terrains, and residential areas. This comprehensive data collection provides vital ground truth for model training. A set of channel features was engineered, including geographical attributes derived from LiDAR datasets. These features were then used to train our prediction model, incorporating the highly efficient and robust gradient boosting ML algorithm, CatBoost. The integration of synthetic data, as demonstrated in our study, significantly improves the generalizability of the model in different environments, achieving a remarkable improvement of approximately 12dB in terms of mean absolute error for the best-case scenario. Moreover, our analysis reveals that even a small fraction of measurements added to the simulation training set, with proper data balance, can significantly enhance the model's performance., Comment: 6 pages, 5 figures, Accepted at ICC 2024

Pancreatic ductal adenocarcinoma (PDAC) presents a critical global health challenge, and early detection is crucial for improving the 5-year survival rate. Recent medical imaging and computational algorithm advances offer potential solutions for early diagnosis. Deep learning, particularly in the form of convolutional neural networks (CNNs), has demonstrated success in medical image analysis tasks, including classification and segmentation. However, the limited availability of clinical data for training purposes continues to provide a significant obstacle. Data augmentation, generative adversarial networks (GANs), and cross-validation are potential techniques to address this limitation and improve model performance, but effective solutions are still rare for 3D PDAC, where contrast is especially poor owing to the high heterogeneity in both tumor and background tissues. In this study, we developed a new GAN-based model, named 3DGAUnet, for generating realistic 3D CT images of PDAC tumors and pancreatic tissue, which can generate the interslice connection data that the existing 2D CT image synthesis models lack. Our innovation is to develop a 3D U-Net architecture for the generator to improve shape and texture learning for PDAC tumors and pancreatic tissue. Our approach offers a promising path to tackle the urgent requirement for creative and synergistic methods to combat PDAC. The development of this GAN-based model has the potential to alleviate data scarcity issues, elevate the quality of synthesized data, and thereby facilitate the progression of deep learning models to enhance the accuracy and early detection of PDAC tumors, which could profoundly impact patient outcomes. Furthermore, this model has the potential to be adapted to other types of solid tumors, hence making significant contributions to the field of medical imaging in terms of image processing models.

Objective: To evaluate the efficiency of large language models (LLMs) such as ChatGPT to assist in diagnosing neuro-ophthalmic diseases based on detailed case descriptions. Methods: We selected 22 different case reports of neuro-ophthalmic diseases from a publicly available online database. These cases included a wide range of chronic and acute diseases that are commonly seen by neuro-ophthalmic sub-specialists. We inserted the text from each case as a new prompt into both ChatGPT v3.5 and ChatGPT Plus v4.0 and asked for the most probable diagnosis. We then presented the exact information to two neuro-ophthalmologists and recorded their diagnoses followed by comparison to responses from both versions of ChatGPT. Results: ChatGPT v3.5, ChatGPT Plus v4.0, and the two neuro-ophthalmologists were correct in 13 (59%), 18 (82%), 19 (86%), and 19 (86%) out of 22 cases, respectively. The agreement between the various diagnostic sources were as follows: ChatGPT v3.5 and ChatGPT Plus v4.0, 13 (59%); ChatGPT v3.5 and the first neuro-ophthalmologist, 12 (55%); ChatGPT v3.5 and the second neuro-ophthalmologist, 12 (55%); ChatGPT Plus v4.0 and the first neuro-ophthalmologist, 17 (77%); ChatGPT Plus v4.0 and the second neuro-ophthalmologist, 16 (73%); and first and second neuro-ophthalmologists 17 (17%). Conclusions: The accuracy of ChatGPT v3.5 and ChatGPT Plus v4.0 in diagnosing patients with neuro-ophthalmic diseases was 59% and 82%, respectively. With further development, ChatGPT Plus v4.0 may have potential to be used in clinical care settings to assist clinicians in providing quick, accurate diagnoses of patients in neuro-ophthalmology. The applicability of using LLMs like ChatGPT in clinical settings that lack access to subspeciality trained neuro-ophthalmologists deserves further research.

Shear flows cause aspherical colloidal particles to tumble so that their orientations trace out complex trajectories known as Jeffery orbits. The Jeffery orbit of a prolate ellipsoid is predicted to align the particle's principal axis preferentially in the plane transverse to the axis of shear. Holographic microscopy measurements reveal instead that colloidal ellipsoids' trajectories in Poiseuille flows strongly favor an orientation inclined by roughly $\pi/8$ relative to this plane. This anomalous observation is consistent with at least two previous reports of colloidal rods and dimers of colloidal spheres in Poiseuille flow and therefore appears to be a generic, yet unexplained feature of colloidal transport at low Reynolds numbers., Comment: 5 pages, 4 figures

We describe a set of novel methods for efficiently sampling high-dimensional parameter spaces of physical theories defined at high energies, but constrained by experimental measurements made at lower energies. Often, theoretical models such as supersymmetry are defined by many parameters, $\mathcal{O}(10-100)$, expressed at high energies, while relevant experimental constraints are often defined at much lower energies, preventing them from directly ruling out portions of the space. Instead, the low-energy constraints define a complex, potentially non-contiguous subspace of the theory parameters. Naive scanning of the theory space for points which satisfy the low-energy constraints is hopelessly inefficient due to the high dimensionality, and the inverse problem is considered intractable. As a result, many theoretical spaces remain under-explored. We introduce a class of modified generative autoencoders, which attack this problem by mapping the high-dimensional parameter space to a structured low-dimensional latent space, allowing for easy visualization and efficient generation of theory points which satisfy experimental constraints. An extension without dimensional compression, which focuses on limiting potential information loss, is also introduced., Comment: Expanded on motivation for methods, added additional metrics and further discussion/comparison of methods

We demonstrate an important step towards on chip integration of single photon sources operating at room temperature fiber coupling of a directional quantum emitter with back-excitation. Directionality is achieved with a hybrid metal-dielectric bullseye antenna, while back-excitation is permitted by placement of the emitter at or in a sub-wavelength hole positioned at the bullseye center. Overall, the unique design enables a direct laser excitation from the back of the on-chip device and very efficient coupling of the highly collimated photon emission to either low numerical aperture (NA) free space optics or directly to an optical fiber from the front. To show the versatility of the concept, we fabricate devices containing either a colloidal quantum dot or a silicon-vacancy center containing nanodiamond, which are accurately coupled to the nano-antenna using two different nano-positioning methods. Both back-excited devices display front collection efficiencies of about 70 % at NAs as low as 0.5. Moreover, the combination of back-excitation with forward low-NA directionality enables direct coupling of the emitted photons into a proximal optical fiber without the need of any coupling optics, thereby facilitating and greatly simplifying future integration.