Detecting robotic traffic at scale on online ads needs an approach that is scalable, comprehensive, precise, and can rapidly respond to changing traffic patterns. In this paper we describe SLIDR or SLIce-Level Detection of Robots, a real-time deep neural network model trained with weak supervision to identify invalid clicks on online ads. We ensure fairness across different traffic slices by formulating a convex optimization problem that allows SLIDR to achieve optimal performance on individual traffic slices with a budget on overall false positives. SLIDR has been deployed since 2021 and safeguards advertiser campaigns on Amazon against robots clicking on ads on the e-commerce site. We describe some of the important lessons learned by deploying SLIDR that include guardrails that prevent updates of anomalous models and disaster recovery mechanisms to mitigate or correct decisions made by a faulty model.

We introduce a discrete network corresponding to any Gaussian wireless network that is obtained by simply quantizing the received signals and restricting the transmitted signals to a finite precision. Since signals in the discrete network are obtained from those of a Gaussian network, the Gaussian network can be operated on the quantization-based digital interface defined by the discrete network. We prove that this digital interface is near optimal for Gaussian relay networks and the capacities of the Gaussian and the discrete networks are within a bounded gap ofO(M2) bits, whereMis the number of nodes. We also prove that any near-optimal coding strategy for the discrete network can be naturally transformed into a near-optimal coding strategy for the Gaussian network merely by quantization. We exploit this property by designing a linear coding strategy for the case of layered discrete relay networks. The linear coding strategy is near optimal and achieves all rates withinO(M2) bits of the capacity, independent of channel gains or signal-to-noise ratio. The linear code is therefore a near-optimal strategy for layered Gaussian relay networks and can be used as on the Gaussian network after simply quantizing the signals. The relays in the linear code need not know the channel gains on either the incoming or the outgoing links. The transmit and receive signals at all relays are simply quantized to binary tuples of the same lengthn, which is all that the nodes need to know. The linear network code is a particularly simple scheme and requires all the relay nodes to collect the received binary tuples into a long binary vector and apply a linear transformation on the long vector. The resulting binary vector is split into smaller binary tuples for transmission by the relays. The quantization requirements of the linear network code are completely defined by the parametern, which therefore also determines the resolution of the analog-to-digital and digital-to-analog converters that are required for operating the network within a bounded gap of the network's capacity. As evident from the description, the linear network code explicitly connects network coding for wireline networks with codes for Gaussian networks.

The cellular industry is evaluating architectures to distribute the signal processing in radio access networks. One of the options is to process the signals of all base stations on a shared pool of compute resources in a central location. In this centralized architecture, the existing base stations will be replaced with just the antennas and a few other active RF components, and the remainder of the digital processing including the physical layer will be carried out in a central location. This model has potential benefits that include a reduction in the cost of operating the network due to fewer site visits, easy upgrades, and lower site lease costs, and an improvement in the network performance with joint signal processing techniques that span multiple base stations. Further there is a potential to exploit variations in the processing load across base stations, to pool the base stations into fewer compute resources, thereby allowing the operator to either reduce energy consumption by turning the remaining processors off or reducing costs by provisioning fewer compute resources. We focus on this aspect in this paper.