Input Mode Research Articles

Fluid Grey 2: How Well Does Generative Adversarial Networks Learn Deeper Topology Structure in Architecture That Matches Images?

Taking into account the regional characteristics of ‘intrinsic’ and ‘extrinsic’ properties of space is an essential issue in architectural design and urban renewal, which is often achieved step by step using image and graph-based GANs. However, each model nesting and data conversion may cause information loss, and it is necessary to streamline the tools to facilitate architects and users to participate in the design. Therefore, this study hopes to prove that I2I GAN also has the potential to recognize topological relationships autonomously. Therefore, this research proposes a method for quickly detecting the ability of pix2pix to learn topological relationships, which is achieved by adding two Grasshopper-based detection modules before and after GAN. At the same time, quantitative data is provided and its learning process is visualized, and changes in different input modes such as greyscale and RGB affect its learning efficiency. There are two innovations in this paper: 1) It proves that pix2pix can automatically learn spatial topological relationships and apply them to architectural design. 2) It fills the gap in detecting the performance of Image-based Generation GAN from a topological perspective. Moreover, the detection method proposed in this study takes a short time and is simple to operate. The two detection modules can be widely used for customizing image datasets with the same topological structure and for batch detection of topological relationships of images. In the future, this paper may provide a theoretical foundation and data support for the application of architectural design and urban renewal that use GAN to preserve spatial topological characteristics.

Aircraft human‐machine interaction assistant design: A novel multimodal data processing and application framework

AbstractDuring aircraft operations, pilots rely on human‐machine interaction platforms to access essential information services. However, the development of a highly usable aerial assistant necessitates the incorporation of two interaction modes: active‐command and passive‐response modes, along with three input modes: voice inputs, situation inputs, and plan inputs. This research focuses on the design of an aircraft human‐machine interaction assistant (AHMIA), which serves as a multimodal data processing and application framework for human‐to‐machine interaction in a fully voice‐controlled manner. For the voice mode, a finetuned FunASR model is employed, leveraging private aeronautical datasets to enable specific aeronautical speech recognition. For the situation mode, a hierarchical situation events extraction model is proposed, facilitating the utilization of high‐level situational features. For the plan mode, a multi‐formations double‐code network plan diagram with a timeline is utilized to effectively represent plan information. Notably, to bridge the gap between human language and machine language, a hierarchical knowledge engine named process‐event‐condition‐order‐skill (PECOS) is introduced. PECOS provides three distinct products: the PECOS model, the PECOS state chart, and the PECOS knowledge description. Simulation results within the air confrontation scenario demonstrate that AHMIA enables active‐command and passive‐response interactions with pilots, thereby enhancing the overall interaction modality.

Comparative Evaluation of Methane Yield from Continuous and Batch Fermentative Processing of Selected Agrowaste

This study investigated the influence of different agrowaste compositions on methane yield in batch and continuous operational modes, highlighting the significance of substrate characteristics in determining the efficiency of methane production. It comparatively assessed the methane yield derived from continuous and batch fermentative processing of the following agrowaste; Cassava peels (A), Maize husks (B), Pig Slurry (C), and composite (D) that represents the co-digestion of the aforementioned A, B & C. Eight digesters were used in all, two anaerobic systems for each substrate, [I.e one operating in a batch mode (1) and the other in a continuous mode (2)]. The proximate properties of the substrates were determined and gas production rates were monitored over the experimental period of 45 days. Gas analysis of the generated gas revealed mainly methane and CO2, with concomitant organic compounds like; Ethanol, Methanol, Acetic Acid, Acetone, etc. Minute quantities of Phenol and Lactic Acid were mainly found in the gas generated from Cassava peels. The average percentage value of Co-digested substrates (D) yielded higher gas volume of 78.263±4.54, followed by C (74.263±4.65), B (70.240±4.38), and A (56.090±3.50). The results also revealed that the continuous fermentative process demonstrated a more stable and consistent methane production rate compared to the batch system, attributed to its steady-state operational mode and continuous substrate feed input. However, modeled graphical results showed that batch fermentation system produced higher gas yield (for all the substrates) between days 29-39 before it recorded a sharp decline curve, attributed to the reduction in substrate availability.

Sequentially Activated-Dumbbell DNA Nanodevices for Accurate Detection of Uracil-DNA Glycosylase via PER-Based Orthogonal Signal Outputs.

Accurate and reliable detection of uracil-DNA glycosylase (UDG) activity is crucial for clinical diagnosis and prognosis assessment. However, current techniques for accurately monitoring UDG activity still face significant challenges due to the single input or output signal modes. Here, we develop a sequentially activated-dumbbell DNA nanodevice (SEAD) that enables precise and reliable evaluation of UDG activity through primer exchange reactions (PER)-based orthogonal signal output. The SEAD incorporates a double-hairpin structure with a stem containing two deoxyuridine (dU) sites for target recognition and two preblocked primer binding regions for target amplification and signal output. Upon UDG recognition of dU, the SEAD can be cleaved by apurinic/apyrimidinic endonuclease 1 (APE1), generating two different hairpins with exposed primer binding regions. These hairpins serve as templates to initiate the parallel PER, enabling the extending of two different amplification products: a long single-stranded DNA (ssDNA) with repetitive sequences and a short ferrocene-labeled ssDNA with complementary sequences. These products further self-assemble into DNA nano-strings in an orthogonal manner that act as an electrochemiluminescence signal switch, enabling precise detection of low-abundance UDG. This work develops a sequential input and orthogonal output strategy for accurately monitoring UDG activity, highlighting the significant potential in cancer diagnosis and treatment.

Deep learning model using planar whole-body bone scintigraphy for diagnosis of skull base invasion in patients with nasopharyngeal carcinoma

PurposeThis study assesses the reliability of deep learning models based on planar whole-body bone scintigraphy for diagnosing Skull base invasion (SBI) in nasopharyngeal carcinoma (NPC) patients.MethodsIn this multicenter study, a deep learning model was developed using data from one center with a 7:3 allocation to training and internal test sets, to diagnose SBI in patients newly diagnosed with NPC using planar whole-body bone scintigraphy. Patients were diagnosed based on a composite reference standard incorporating radiologic and follow-up data. Ten different convolutional neural network (CNN) models were applied to both whole-image and partial-image input modes to determine the optimal model for each analysis. Model performance was assessed using the area under the receiver operating characteristic curve (AUC), calibration, decision curve analysis (DCA), and compared with expert assessments by two nuclear medicine physicians.ResultsThe best-performing model using partial-body input achieved AUCs of 0.80 (95% CI: 0.73, 0.86) in the internal test set, 0.84 (95% CI: 0.77, 0.91) in the external cohort, and 0.78 (95% CI: 0.73, 0.83) in the treatment test cohort. Calibration curves and DCA confirmed the models’ excellent discrimination, calibration, and potential clinical utility across internal and external datasets. The AUCs of both nuclear medicine physicians were lower than those of the best-performing deep learning model in external test set (AUC: 0.75 vs. 0.77 vs. 0.84).ConclusionDeep learning models utilizing partial-body input from planar whole-body bone scintigraphy demonstrate high discriminatory power for diagnosing SBI in NPC patients, surpassing experienced nuclear medicine physicians.

Synthetic Image Generation Using Deep Learning: A Systematic Literature Review

ABSTRACTThe advent of deep neural networks and improved computational power have brought a revolutionary transformation in the fields of computer vision and image processing. Within the realm of computer vision, there has been a significant interest in the area of synthetic image generation, which is a creative side of AI. Many researchers have introduced innovative methods to identify deep neural network‐based architectures involved in image generation via different modes of input, like text, scene graph layouts and so forth to generate synthetic images. Computer‐generated images have been found to contribute a lot to the training of different machine and deep‐learning models. Nonetheless, we have observed an immediate need for a comprehensive and systematic literature review that encompasses a summary and critical evaluation of current primary studies' approaches toward image generation. To address this, we carried out a systematic literature review on synthetic image generation approaches published from 2018 to February 2023. Moreover, we have conducted a systematic review of various datasets, approaches to image generation, performance metrics for existing methods, and a brief experimental comparison of DCGAN (deep convolutional generative adversarial network) and cGAN (conditional generative adversarial network) in the context of image generation. Additionally, we have identified applications related to image generation models with critical evaluation of the primary studies on the subject matter. Finally, we present some future research directions to further contribute to the field of image generation using deep neural networks.

Reconfigurable mode converter based on a Sb2Se3 phase change material and inverse design

In this paper, we combine the inverse design with a silicon-Sb2Se3 hybrid platform to design an on-chip mode converter that converts basic modes to higher-order modes. Firstly, we present a 1 × 2 mode converter with dimensions of 4.8 × 2.7 µm2 that enables TE0 mode input, TE0 or TE1 output in the C-band (1530 nm to 1565 nm) with an insertion loss (IL) of less than 0.8 dB and a crosstalk (CT) of less than -13 dB. Secondly, the device is extended to a 1 × 3 switchable three-mode converter. Using two controllable phase change regions as drivers, it can flexibly control the switching from TE0 mode input to three modes of TE0, TE1, or TE2 outputs, which enables mode switching and signal routing. The device can be switched between three modes and has broad application potential in broadband optical signal processing for mode division multiplexing systems, as well as optical interconnections. Finally, the device is extended to a 1 × 2 controllable (mode and power) beam splitter, which can control the power ratio between output modes. By modulating the crystallinity of Sb2Se3, the simulation achieves a multilevel switching of 36 levels (> 5-bit). These devices pave the way for high integration densities in future photonic chips.

Performance of Publicly Available Large Language Models on Internal Medicine Board-style Questions.

Ongoing research attempts to benchmark large language models (LLM) against physicians' fund of knowledge by assessing LLM performance on medical examinations. No prior study has assessed LLM performance on internal medicine (IM) board examination questions. Limited data exists on how knowledge supplied to the models, derived from medical texts improves LLM performance. The performance of GPT-3.5, GPT-4.0, LaMDA and Llama 2, with and without additional model input augmentation, was assessed on 240 randomly selected IM board-style questions. Questions were sourced from the Medical Knowledge Self-Assessment Program released by the American College of Physicians with each question serving as part of the LLM prompt. When available, LLMs were accessed both through their application programming interface (API) and their corresponding chatbot. Mode inputs were augmented with Harrison's Principles of Internal Medicine using the method of Retrieval Augmented Generation. LLM-generated explanations to 25 correctly answered questions were presented in a blinded fashion alongside the MKSAP explanation to an IM board-certified physician tasked with selecting the human generated response. GPT-4.0, accessed either through Bing Chat or its API, scored 77.5-80.7% outperforming GPT-3.5, human respondents, LaMDA and Llama 2 in that order. GPT-4.0 outperformed human MKSAP users on every tested IM subject with its highest and lowest percentile scores in Infectious Disease (80th) and Rheumatology (99.7th), respectively. There is a 3.2-5.3% decrease in performance of both GPT-3.5 and GPT-4.0 when accessing the LLM through its API instead of its online chatbot. There is 4.5-7.5% increase in performance of both GPT-3.5 and GPT-4.0 accessed through their APIs after additional input augmentation. The blinded reviewer correctly identified the human generated MKSAP response in 72% of the 25-question sample set. GPT-4.0 performed best on IM board-style questions outperforming human respondents. Augmenting with domain-specific information improved performance rendering Retrieval Augmented Generation a possible technique for improving accuracy in medical examination LLM responses.

Detection model of atrial fibrillation based on multi-branch and multi-scale convolutional networks

Atrial fibrillation (AF) is a life-threatening heart condition, and its early detection and treatment have garnered significant attention from physicians in recent years. Traditional methods of detecting AF heavily rely on doctor's diagnosis based on electrocardiograms (ECGs), but prolonged analysis of ECG signals is very time-consuming. This paper designs an AF detection model based on the Inception module, constructing multi-branch detection channels to process raw ECG signals, gradient signals, and frequency signals during AF. The model efficiently extracted QRS complex and RR interval features using gradient signals, extracted P-wave and f-wave features using frequency signals, and used raw signals to supplement missing information. The multi-scale convolutional kernels in the Inception module provided various receptive fields and performed comprehensive analysis of the multi-branch results, enabling early AF detection. Compared to current machine learning algorithms that use only RR interval and heart rate variability features, the proposed algorithm additionally employed frequency features, making fuller use of the information within the signals. For deep learning methods using raw and frequency signals, this paper introduced an enhanced method for the QRS complex, allowing the network to extract features more effectively. By using a multi-branch input mode, the model comprehensively considered irregular RR intervals and P-wave and f-wave features in AF. Testing on the MIT-BIH AF database showed that the inter-patient detection accuracy was 96.89%, sensitivity was 97.72%, and specificity was 95.88%. The proposed model demonstrates excellent performance and can achieve automatic AF detection.

Convergence analysis of a novel high order networks model based on entropy error function

It is generally known that the error function is one of the key factors that determine the convergence, stability and generalization ability of neural networks. For most feedforward neural networks, the squared error function is usually chosen as the error function to train the network. However, networks based on the squared error function can lead to slow convergence and easily fall into local optimum in the actual training process. Recent studies have found that, compared to the squared error function, the gradient method based on the entropy error function measures the difference between the probability distribution of the model output and the probability distribution of the true labels during the iterative process, which can be more able to handle the uncertainty in the classification problem, less likely to fall into a local optimum and can learn to converge more rapidly. In this paper, we propose a batch gradient method for Sigma-Pi-Sigma neural networks based on the entropy error function and rigorously demonstrate the weak and strong convergence of the new algorithm in the batch input mode. Finally, the theoretical results and effectiveness of the algorithm are verified by simulation.

Polarization Analysis of Vertically Etched Lithium Niobate-on-Insulator (LNOI) Devices

LNOI devices have emerged as prominent contributors to photonic integrated circuits (PICs), benefiting from their outstanding performance in electro-optics, acousto-optics, nonlinear optics, etc. Due to the physical properties and current etching technologies of LiNbO3, slanted sidewalls are typically formed in LNOI waveguides, causing polarization-related mode hybridization and crosstalk. Despite the low losses achieved with fabrication advancements in LNOI, such mode hybridization and crosstalk still significantly limit the device performance by introducing polarization-related losses. In this paper, we propose a vertically etched LNOI construction. By improving the geometrical symmetry in the waveguides, vertical sidewalls could adequately mitigate mode hybridization in common waveguide cross sections. Taking tapers and bends as representatives of PIC components, we then conducted theoretical modeling and simulations, which showed that vertical etching effectively exempts devices from polarization-related mode crosstalk. This not only improves the polarization purity and input mode transmittance but also enables lower polarization-related losses within more compact structures. As a demonstration of fabrication feasibility, we innovatively proposed a two-step fabrication technique, and successfully fabricated waveguides with vertical sidewalls. Such vertical etching technology facilitates the development of next-generation high-speed modulators, nonlinear optical devices, and other advanced photonic devices with lower losses and a smaller footprint, driving further innovations in both academic research and industrial applications.

Effects of energy input modes on energy efficiency and berry quality in continuous microwave drying

To elucidate the effects of energy input modes on energy efficiency and berry quality in continuous microwave drying, four energy input modes with variational working ways of magnetrons were investigated by using numerical simulation and bench experiments in a continuous microwave dryer. Energy efficiency was divided into absorption and conversion efficiency to study the change rule of drying process, and the retention rate of anthocyanins of berry’s representative nutrition ingredient was studied considering both of the temperature and its uniformity. Results indicate that the energy input modes affect the absorption efficiency by changing the distribution uniformity of electric field of material layer, and the conversion efficiency by altering the matching degree between the energy allocation and drying characteristics of berry, which are both determined by the ways of working magnetrons of different energy input modes. The conversion efficiency has a greater influence on the final energy efficiency, so the energy input mode should be paid attention to the provision of microwave energy by selecting proper way of working magnetrons according to the drying characteristics. The effect of temperature uniformity on anthocyanin degradation is more obvious with average temperature increasing gradually, where better uniformity of electric field strength should be provided to guarantee the quality of dried berry by adjusting the energy input mode. The research results provide significant insights to optimize the energy input mode of taking into account both the energy efficiency and dried quality of material.

Classical capacity of quantum non-Gaussian attenuator and amplifier channels

We consider a quantum bosonic channel that couples the input mode via a beam splitter or two-mode squeezer to an environmental mode that is prepared in an arbitrary state. We investigate the classical capacity of this channel, which we call a non-Gaussian attenuator or amplifier channel. If the environment state is thermal, we of course recover a Gaussian phase-covariant channel whose classical capacity is well known. Otherwise, we derive both a lower and an upper bound to the classical capacity of the channel, drawing inspiration from the classical treatment of the capacity of non-Gaussian additive-noise channels. We show that the lower bound to the capacity is always achievable and give examples where the non-Gaussianity of the channel can be exploited so that the communication rate beats the capacity of the Gaussian-equivalent channel (i.e. the channel where the environment state is replaced by a Gaussian state with the same covariance matrix). Finally, our upper bound leads us to formulate and investigate conjectures on the input state that minimizes the output entropy of non-Gaussian attenuator or amplifier channels. Solving these conjectures would be a main step toward accessing the capacity of a large class of non-Gaussian bosonic channels.

Spiking mode-based neural networks.

Spiking neural networks play an important role in brainlike neuromorphic computations and in studying working mechanisms of neural circuits. One drawback of training a large-scale spiking neural network is that updating all weights is quite expensive. Furthermore, after training, all information related to the computational task is hidden into the weight matrix, prohibiting us from a transparent understanding of circuit mechanisms. Therefore, in this work, we address these challenges by proposing a spiking mode-based training protocol, where the recurrent weight matrix is explained as a Hopfield-like multiplication of three matrices: input modes, output modes, and a score matrix. The first advantage is that the weight is interpreted by input and output modes and their associated scores characterizing the importance of each decomposition term. The number of modes is thus adjustable, allowing more degrees of freedom for modeling the experimental data. This significantly reduces the training cost because of significantly reduced space complexity for learning. Training spiking networks is thus carried out in the mode-score space. The second advantage is that one can project the high-dimensional neural activity (filtered spike train) in the state space onto the mode space which is typically of a low dimension, e.g., a few modes are sufficient to capture the shape of the underlying neural manifolds. We successfully apply our framework in two computational tasks-digit classification and selective sensory integration tasks. Our method thus accelerates the training of spiking neural networks by a Hopfield-like decomposition, and moreover this training leads to low-dimensional attractor structures of high-dimensional neural dynamics.

Incidental Learning of Collocations Under Different Input Modes and the Mediating Role of Perceptual Learning Style.

This study investigated how input modes (reading vs. listening) and learners' perceptual learning style (visual vs. auditory) affected the incidental learning of collocations. A total of 182 college students were first assigned to either a visual or auditory group based on their performance on a perceptual learning style questionnaire. Each style group was subsequently subdivided into three groups who were exposed to a series of texts containing unfamiliar collocation items under one of the input conditions: written input, aural input, or no input. Results of the study indicated that both written and aural input led to gains in collocational knowledge, and aural input was more effective than written input. Furthermore, the study provided empirical evidence that there was a moderating role of perceptual learning style on incidental collocation learning. The auditory learners under aural input showed the highest rate of collocation learning among all treatment subgroups.

Incidental learning of collocations through different multimodal input: The role of learners’ initial L2 proficiency

This study investigated the impact of one unimodal (i.e., reading-only) and two different multimodal input modes (i.e., reading-while-listening and viewing with captions) on incidental collocation learning, alongside examining the relationship between learning gains and overall L2 proficiency. A total of 180 university students at pre-intermediate and upper-intermediate levels from China were randomly assigned to one of the input modes. Target collocation knowledge tests, including form recognition and meaning recall, were administered immediately after exposure to the input material and after a 2-week interval. Results indicated that while learners across different proficiency levels acquired collocation knowledge through all input modes to varying degrees, those in the multimodal conditions outperformed their counterparts in terms of both recognition and recall dimensions. Additionally, after 2 weeks, despite decreased learning gains for all learner groups, multimodal input was found to benefit upper-intermediate learners' meaning recall of the target collocations. Learners’ language proficiency significantly predicted their immediate learning outcomes in both recognition and recall. These findings offer further support for multimedia learning theory and dual code theory, contributing to a deeper understanding of the value of multimodal meaning-focused language input.

Perfect pulsed inline twin-beam squeezers

Perfect inline squeezers are both spectrally pure and have identical input and output temporal modes, allowing one to squeeze an arbitrary input quantum state in the sole input mode on which the device acts, while the quantum states of any other modes are unaffected. We study theoretically how to obtain a perfect pulsed inline squeezer in twin-beam systems by considering three commonly used configurations: unpoled single pass, poled single pass, and poled double pass. By obtaining analytical relations between the input and output temporal modes from the Bloch–Messiah decomposition of the discretized Heisenberg-picture propagator, we find that a double-pass structure produces a perfect pulsed inline squeezer when operated in a frequency degenerate, symmetric group-velocity matched type-II configuration.

A mechanistic deterioration point assignment model for water pipe condition assessment

ABSTRACT A pipe condition assessment model is required to implement effective and economical planned maintenance of the water distribution system. The application of such a model requires sufficient accuracy, which, however, is limited by the complexity of the pipe deterioration process and data storage capacity of the water utility. The majority of previous studies have focused on the improvement of assessment algorithms for data mining. In this study, a mechanistic deterioration point assignment (MDPA) model is developed to make advancements in the modes of data input and result output to enhance the model's accuracy and application scope for cast iron and steel pipes. In this MDPA model, (1) indicators/sub-indicators on external corrosion, external load, internal corrosion, and internal load are constructed and can be obtained by data estimation or techniques and (2) assessment results include both pipe overall condition and detailed conditions on corrosion and load, offering evidence for primary maintenance measures. The weights of the indicators/sub-indicators are estimated using the Bayesian statistics theory. The modelling results of pipe samples demonstrate that this MDPA model is an effective tool for pipe condition assessment.

An Audio - Visual Virtual Personal Assistant

Current traits in smart assistants and smart home automation are currently attracting the interests of customers and researchers. Speech enabled smart virtual assistants (named smart speakers) offer a wide sort of network orientated services and, in some instances, can connect to smart environments, accordingly improving them with new and effective user interfaces. But, such gadgets also reveal new desires and a few weaknesses. Specially, they constitute faceless and blind assistants, not able to reveal a face, and therefore an emotion, and not able to ‘see’ the person. For this reason, the interaction is impaired and, in a few cases, ineffective. One of the goals of artificial intelligence is the realization of natural talk among humans and machines. In latest years, the dialogue systems, known as interactive conversational systems are the fastest developing vicinity in AI. Many companies have used the dialogue systems technology to set up various sorts of VPAs based on their application areas, including Apple’s Siri, Amazon Alexa, etc. To triumph over such troubles, in this project we combine a number of advance techniques. The proposed Assistant is powerful and resource-efficient, interactive and customizable. We use the multi-modal dialogue structures and screen projection which process two or more combined user input modes, which includes speech, image, video, touch, manual gestures, body movements with the intention to design the NextGeneration of VPAs. The new version of VPAs can be used to increase the interaction between humans and machines by the usage of distinctive technologies, consisting of gesture recognition, image/video recognition, speech recognition, dialogue system, conversational information base, and the overall knowledge base. Furthermore, this VPA device can be utilized in different areas such as education, medical assistance, disabilities systems, home automation and security access control.

Intracellular Pharmacokinetics of Activated Drugs in a Prodrug-Enzyme-Ultrasound System: Evaluations on ZD2767P+CPG2+US.

Intracellular pharmacokinetics (PK) of activated drugs is a window to understanding the pharmacodynamics of prodrug-enzyme-ultrasound therapy. Herein PK of ZD2767D (i.e., activated drug) in the ZD2767P+CPG2+US system on A549, A549/DDP, SKOV3, and SKOV3/DDP cells were evaluated (A549/DDP and SKOV3/DDP were cisplatin-resistant sublines). The noncompartment model under extravascular input mode was deemed appropriate for evaluating drug level vs time curves; Cmax, AUClast, MRTlast, Vz, and Cl can reflect the PK feature, but t1/2, AUCinf, and MRTinf were irrational; higher accumulation and slower elimination characterized the PK mechanism of ZD2767P+CPG2+US; enhanced permeability and retention effect can be assessed with Cmax, AUClast, MRTlast, and tlast; ultrasound equivalently modulated Cmax and AUClast in sensitive and resistant cells. The experimental design and dose proportionality were discussed.