In machine learning tasks, models trained in the source domain often suffer from performance degradation in the target domain due to domain drift or distribution shift. In this paper, we explore the concept of sensor-actuator design in adaptive control to address this domain drift problem and develop a new approach, called learning inference-time drift sensor-actuator (LIDSA) for domain generalization. The drift sensor network consists of a constraint network and a data converter. The constraint network is learned to extract a set of constraints in the source domain and sense the domain drift by detecting the deviation from these constraints, called constraint error, which is correlated with the classification error. The data converter network then maps this constraint error into an effective guidance signal, which can guide the actuator network to adjust the feature to achieve improved discrimination power and better generalization performance. Our extensive experimental results demonstrate that the proposed LIDSA approach improves the performance of domain generalization over the baseline method.

A central challenge in human pose estimation, as well as in many other machine learning and prediction tasks, is the generalization problem. The learned network does not have the capability to characterize the prediction error, generate feedback information from the test sample, and correct the prediction error on the fly for each individual test sample, which results in degraded performance in generalization. In this work, we introduce a self-correctable and adaptable inference (SCAI) method to address the generalization challenge of network prediction and use human pose estimation as an example to demonstrate its effectiveness and performance. We learn a correction network to correct the prediction result conditioned by a fitness feedback error. This feedback error is generated by a learned fitness feedback network which maps the prediction result to the original input domain and compares it against the original input. Interestingly, we find that this self-referential feedback error is highly correlated with the actual prediction error. This strong correlation suggests that we can use this error as feedback to guide the correction process. It can be also used as a loss function to quickly adapt and optimize the correction network during the inference process. Our extensive experimental results on human pose estimation demonstrate that the proposed SCAI method is able to significantly improve the generalization capability and performance of human pose estimation.

Fully test-time adaptation aims to adapt the network model based on sequential analysis of input samples during the inference stage to address the cross-domain performance degradation problem of deep neural networks. We take inspiration from the biological plausibility learning where the neuron responses are tuned based on a local synapse-change procedure and activated by competitive lateral inhibition rules. Based on these feed-forward learning rules, we design a soft Hebbian learning process which provides an unsupervised and effective mechanism for online adaptation. We observe that the performance of this feed-forward Hebbian learning for fully test-time adaptation can be significantly improved by incorporating a feedback neuromodulation layer. It is able to fine-tune the neuron responses based on the external feedback generated by the error backpropagation from the top inference layers. This leads to our proposed neuro-modulated Hebbian learning (NHL) method for fully test-time adaptation. With the unsupervised feed-forward soft Hebbian learning being combined with a learned neuromodulator to capture feedback from external responses, the source model can be effectively adapted during the testing process. Experimental results on benchmark datasets demonstrate that our proposed method can significantly improve the adaptation performance of network models and outperforms existing state-of-the-art methods.

In prior work, we proposed a cross-layer architecture called Multicast-Push Unicast-Pull (MPUP) for Software Defined Networks (SDN) to support a reliable file-stream multicast application. In this work, we improved the algorithms used to set parameters: transport-layer sender retransmission timer, VLAN rate (which is also the sending rate) and sender-buffer size. Experimental evaluation using feeds with metadata collected from real meteorology file streams was conducted. A significant finding is that the throughput achieved is smaller than the VLAN/sending rate even though file blocks are multicast continuously in UDP datagrams. Sender-buffer waiting times and propagation delays are the main reasons for the degraded throughput. For example, increasing the VLAN rate from 20 Mbps to 500 Mbps, reduced the degradation from 90% to 45%. However, the degradation increased from 45% to 58% when the VLAN rate was increased from 500 Mbps to 1 Gbps. We found an increase in the number of block retransmissions at the higher rates, which explains this increased degradation. Increasing RTT from 0.1 ms to 100 ms caused throughput to drop from 274.8 Mbps to 27.6 Mbps on a 500 Mbps VLAN. If transmission delay was a significant component in total latency, then throughput degradation relative to VLAN rate would be small; however, the meteorology file-streams used in our study have small-sized data products. Due to bandwidth borrowing between VLAN and IP-routed services, VLAN utilization is not important, and hence we recommend using the smallest rate at which sender-buffer waiting times are insignificant.

The growing deployment of OpenFlow/SDN networks makes it increasingly possible to leverage network multicast services. This work proposes a novel cross-layer Multicast-Push Unicast Pull (MPUP) architecture that includes functionality in the application, transport and link layers to offer users a reliable file-stream distribution service to multiple subscribers. A prototype implementation of the MPUP architecture was realized in a new version of Local Data Manager (LDM), LDM7, a software program that has been in use since 1994 for real-time meteorology data distribution. LDM6, the currently deployed version, uses application-layer multicast. Experiments were run on the GENI infrastructure to compare LDM7 and LDM6. The two main findings are (i) LDM7 can be run at a higher sending rate than LDM6 allowing for improved performance (lower filedelivery latency), and (ii) to achieve the same performance, LDM7 uses significantly lower bandwidth and compute capacity. A three-fold improvement in performance improvement was possible with LDM7, and a bandwidth reduction from 350 Mbps to 21.4 Mbps was observed with 24 receivers.