Upstream Connections Research Articles

Discrepant Semantic Diffusion Boosts Transfer Learning Robustness

Transfer learning could improve the robustness and generalization of the model, reducing potential privacy and security risks. It operates by fine-tuning a pre-trained model on downstream datasets. This process not only enhances the model’s capacity to acquire generalizable features but also ensures an effective alignment between upstream and downstream knowledge domains. Transfer learning can effectively speed up the model convergence when adapting to novel tasks, thereby leading to the efficient conservation of both data and computational resources. However, existing methods often neglect the discrepant downstream–upstream connections. Instead, they rigidly preserve the upstream information without an adequate regularization of the downstream semantic discrepancy. Consequently, this results in weak generalization, issues with collapsed classification, and an overall inferior performance. The main reason lies in the collapsed downstream–upstream connection due to the mismatched semantic granularity. Therefore, we propose a discrepant semantic diffusion method for transfer learning, which could adjust the mismatched semantic granularity and alleviate the collapsed classification problem to improve the transfer learning performance. Specifically, the proposed framework consists of a Prior-Guided Diffusion for pre-training and a discrepant diffusion for fine-tuning. Firstly, the Prior-Guided Diffusion aims to empower the pre-trained model with the semantic-diffusion ability. This is achieved through a semantic prior, which consequently provides a more robust pre-trained model for downstream classification. Secondly, the discrepant diffusion focuses on encouraging semantic diffusion. Its design intends to avoid the unwanted semantic centralization, which often causes the collapsed classification. Furthermore, it is constrained by the semantic discrepancy, serving to elevate the downstream discrimination capabilities. Extensive experiments on eight prevalent downstream classification datasets confirm that our method can outperform a number of state-of-the-art approaches, especially for fine-grained datasets or datasets dissimilar to upstream data (e.g., 3.75% improvement for Cars dataset and 1.79% improvement for SUN dataset under the few-shot setting with 15% data). Furthermore, the experiments of data sparsity caused by privacy protection successfully validate our proposed method’s effectiveness in the field of artificial intelligence security.

Dual-Polarized WDM Access Network With Fiber to the Extension (FTTE) Connection

In this study, a symmetric polarization-division-multiplexing (PDM) wavelength-division-multiplexing passive optical network (WDM-PON) with fiber to the extension (FTTE) architecture is presented and investigated. The proposed PON can also avoid the Rayleigh backscattering (RB) interference noise. Here, each WDM wavelength can produce 25 Gbit/s on-off keying (OOK) and 10 Gbit/s OOK modulation channels simultaneously in x- and y-polarizations (Pol <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">x</sub> and Pol <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">y</sub> ) through PDM method for 25 and 105 km single-mode fiber (SMF) transmissions without dispersion compensation. The measured minimal power budget of 25 and 10 Gbit/s downstream and upstream signals are 29 and 35.1 dB and 29.8 and 31.5 dB, respectively, at the forward error correction (FEC) target. Therefore, the proposed PDM WDM-PON architecture can not only offer symmetric 35 Gbit/s downstream and upstream connections by the dual-polarized wavelength, but also can avoid the RB-induced interference noise. In addition, the presented dual-polarized downstream signal also can provide information security in physical layer, when the unauthorized user occupies unlawfully.

New surgical paradigm for open neural tube defects.

An open neural tube defect (ONTD) features an exposed, unclosed neural plate in the form of an expanded, flat, and frequently hefty neural placode. Traditional philosophy of ONTD repair aims at preserving function at any cost, which often means stuffing the entire thick and unwieldy but non-functional placode into a tight dural sac, increasing the likelihood of future tethering of the spinal cord. The same philosophy of attempting to save the whole perimetry of the placode also sometimes leads to inadvertent inclusion of parts of the squamous epithelial membrane surrounding the placode into the reconstructed product, only to form inclusion dermoid cyst causing further injury to the neural tissues. Lastly, defective closure of the caudal primary neural tube usually results in abolition of secondary and junctional neurulation, leaving a defective conus and sacral nerve roots, clinically presenting in most cases with neurogenic bladder and bowel dysfunction. Preserving this trapped but locally active sacral micturition center, isolated from suprasegmental inhibitory moderation, leads to a spastic, hyperactive, low compliance, and high-pressure bladder predisposing to upstream kidney damage, without benefits of normal bladder function. We report the post-natal surgical treatment of 8 newborn infants with ONTD, in which we resected the non-functional portion of the neural placode identified as such by direct spinal cord/placode and nerve root stimulation, as well as by transcortical evoked motor responses to check for suprasegmental corticospinal connectivity. Any part of the placode without local function or upstream connections was resected, and the small caudal spinal cord stump closed with pia-to-pia microsutures. The patients were followed for pre- and post-operative neuro-urological status and with serial magnetic resonance imaging (MRI) at 3weeks, 6months, and 2years post-repair. Follow-up period ranged from birth to 3years (mean of 24months). Of the 8 patients analyzed, 7 had a terminal and one a segmental ONTD. Pre-operative neurological level ranged from L4 to S2. Applying our new surgical paradigm, we found no neurological worsening post-operatively. All patients had a neurogenic bladder and bowel dysfunction but none had a high-pressure bladder on urodynamics studies. Early and late MRIs all showed a loose and capacious neural placode to dural sac relationship. None had an inclusion dermoid cyst. We propose a new paradigm for the surgical repair of open neural tube defects with intraoperative neuromonitoring and introduce a safe and reliable technique of placode debulking.

Integration of FSO Traffic in Ring-Topology Bidirectional Fiber Access Network With Fault Protection

In this work, we first propose and demonstrate a bidirectional time-division-multiplexing (TDM) free space optical communication (FSO) traffic integrated in ring-based fiber access network together with fiber fault protection. Here, while we add the new designed optical devices in the central office (CO) and each optical wireless unit (OWU) respectively, the TDM-FSO fiber system can achieve the self-protected operation against fault. In addition, according to the proposed ring-based fiber network, the maximum numbers of eight OWUs can be provided for the symmetric 10 Gbps on-off keying (OOK) FSO downstream and upstream connections. As a result, the proposed FSO access network can allow 25 km fiber transmission and proper FSO transmission length depending on the atmospheric attenuation of different weather condition.

Leveraging who you know by what you know: Specialization and returns to relational capital

Research Summary:This paper investigates the interaction effects of specialization and relational capital on performance. We distinguish between upstream and downstream relational capital and theorize that higher levels of specialization will buffer against decreases in upstream relational capital, because of deeper domain expertise and stronger downstream relational capital. Conversely, higher levels of generalization permit greater gains from increases in upstream relational capital, due to leverage across a more diversified downstream portfolio of activities. We test and find support for these hypotheses in the context of the US lobbying industry. Our study contributes to the strategic human capital literature by isolating the dimension of specialization and relational capital embodied within individuals and providing performance implications of the interactions.Managerial Summary:Both “what you know” and “whom you know” impacts performance. Generalists and specialists are different on the “what you know” dimension. On the “who you know” dimension, we distinguish between upstream (supplier) and downstream (client) relationships. We show that specialists are buffered by deeper downstream relations from performance declines when their powerful upstream connections lose power. Generalists benefit from broader networks when their upstream connections gain power. Thus, when the value of their relationships change, specialists and generalists should each assess when they can reap performance benefits, and when they need to bolster against adversities. For firms, our study suggests hiring the right mix of specialists and generalists is important to reduce risks from relational losses while enjoying the performance benefits from relational gains.

Evolution of abandoned channels: Insights on controlling factors in a multi-pressure river system

In the second half of the 19th century, channelization of large multi-thread rivers such as the Rhine, the Danube, and the Rhône rivers induced artificial disconnection of most of their secondary channels. Compared to naturally abandoned channels, terrestrialization (i.e., the passage from the aquatic to the terrestrial stage, controlled by sediment deposits and/or lowering of the water level) patterns and rates of such artificially prematurely abandoned channels remain largely unknown. Moreover, factors controlling their evolutionary trajectories are complex owing to a set of pressures occurring throughout the 20th century within specific space-time windows. Through a case study of the Rhône River, this paper aims to assess and distinguish the effects of a set of potential controlling factors on abandoned channel terrestrialization dynamics and lifespan. We tested the influence of: (i) submersible embankments closing the entrance of abandoned channels, (ii) main channel degradation following its channelization or the water level lowering due to channel bypassing in the middle of the 20th century involving drastic water abstraction in these reaches, (iii) transverse dykes within the abandoned channels, (iv) the flooding regime of abandoned channels (i.e., frequency and magnitude of upstream connections producing lotic functioning), and (v) longitudinal variation in the suspended sediment concentration along the main channel. To this end, we studied 24 abandoned channels (16 artificially disconnected at their upstream end by submersible embankments and eight naturally disconnected by bar plug establishment) from the mid-19th to the beginning of the 20th century. Their terrestrialization rates were characterized through the reconstruction of their planimetric trajectories using historical maps and aerial photos.The results reveal a much longer lifespan of artificial abandoned channels compared to natural ones because of the truncation of the initial bedload infilling phase due to the artificial and imposed closing of their entrance. Moreover, terrestrialization occurred faster when water level lowering or channel degradation was greater. Surprisingly, terrestrialization rates were the highest in the most frequently connected artificial abandoned channels (i.e., channels with a high frequency of lotic functioning), probably in relation to the roughness induced by the presence of transversal dykes. Finally, it is difficult to rank all the factors tested because of their complex combinations, which can change in space and time.

Cerebellar Nuclear Neurons Use Time and Rate Coding to Transmit Purkinje Neuron Pauses.

Neurons of the cerebellar nuclei convey the final output of the cerebellum to their targets in various parts of the brain. Within the cerebellum their direct upstream connections originate from inhibitory Purkinje neurons. Purkinje neurons have a complex firing pattern of regular spikes interrupted by intermittent pauses of variable length. How can the cerebellar nucleus process this complex input pattern? In this modeling study, we investigate different forms of Purkinje neuron simple spike pause synchrony and its influence on candidate coding strategies in the cerebellar nuclei. That is, we investigate how different alignments of synchronous pauses in synthetic Purkinje neuron spike trains affect either time-locking or rate-changes in the downstream nuclei. We find that Purkinje neuron synchrony is mainly represented by changes in the firing rate of cerebellar nuclei neurons. Pause beginning synchronization produced a unique effect on nuclei neuron firing, while the effect of pause ending and pause overlapping synchronization could not be distinguished from each other. Pause beginning synchronization produced better time-locking of nuclear neurons for short length pauses. We also characterize the effect of pause length and spike jitter on the nuclear neuron firing. Additionally, we find that the rate of rebound responses in nuclear neurons after a synchronous pause is controlled by the firing rate of Purkinje neurons preceding it.

Carbon and water cycling in lake‐rich landscapes: Landscape connections, lake hydrology, and biogeochemistry

Lakes are low‐lying connectors of uplands and wetlands, surface water and groundwater, and though they are often studied as independent ecosystems, they function within complex landscapes. One such highly connected region is the Northern Highland Lake District (NHLD), where more than 7000 lakes and their watersheds cycle water and carbon through mixed forests, wetlands, and groundwater systems. Using a new spatially explicit simulation framework representing these coupled cycles, the Lake, Uplands, Wetlands Integrator (LUWI) model, we address basic regional questions in a 72‐lake simulation: (1) How do simulated water and carbon budgets compare with observations, and what are the implications for carbon stocks and fluxes? (2) How do the strength and spatial pattern of landscape connections vary among watersheds? (3) What is the role of interwatershed connections in lake carbon processing? Results closely coincide with observations at seasonal and annual scales and indicate that the connections among components and watersheds are critical to understanding the region. Carbon and water budgets vary widely, even among nearby lakes, and are not easily predictable using heuristics of lake or watershed size. Connections within and among watersheds exert a complex, varied influence on these processes: Whereas inorganic carbon budgets are strongly related to the number and nature of upstream connections, most organic lake carbon originates within the watershed surrounding each lake. This explicit incorporation of terrestrial and aquatic processes in surface and subsurface connection networks will aid our understanding of the relative roles of on‐land, in‐lake, and between‐lake processes in this lake‐rich region.

Tracing of the Stream Flowing Through the Cave Ferranova Buža, Central Slovenia

The article discusses the characteristics of the recently discovered ground water stream in the cave Ferranova buža and its possible karstic hydrological system. Its upstream connections have been narrowed to few possible springs by means of logical exclusion and have yet to be confirmed. For the purpose of establishing its linkage to the downstream springs around Vrhnika a water tracing experiment has been performed. Uranine was used as a tracer and sampled in intervals of eight hours at Primc’s spring, Bečkaj’s spring and Kožuh’s spring. The connection was proved in all three observation points, with Kožuh’s spring draining the largest quantity of the tracer. The tracer experiment evidenced that the dye was completely flushed downward in single transition curve almost without any retention.

Species immigration, extinction and turnover of vascular plants in boreal lakes

Dictated by limited resource availability for land acquisition, a central question in conservation biology is the ability of areas of different size to maintain species diversity. The selected reserves should not only be species rich at the moment, but should also maintain species diversity in the long run. We used two sets of data on vascular plant species in boreal lakes collected in 1933/34 and 1996 to test the relationships between lake area and the extinction, immigration and turnover rates of the species. Moreover, we investigated, whether the number of species in 1933/34 or water connection between lakes was related to extinction, immigration and turnover rates of species. We found that lake area or shoreline length was not correlated with immigration or turnover rate. But extinction rate was slightly negatively correlated with shoreline length. The original number of species was positively related to the number of species extinctions and to the absolute turnover rate in the lakes, which indicates that species richness does not create stability in these communities. Species number was not correlated with immigration rate. Upstream water connections in the lakes did not affect immigration, extinction or turnover rates. We conclude that length of the shoreline is a better measure of suitable area for water plants than the lake area, and that because the correlation between shoreline length and extinction rate was slight, also small lakes can be valuable for conservation.