Digital computing-in-memory (DCIM) enables efficient deep neural networks (DNNs) acceleration but faces limitations in resource overhead, energy efficiency, and architectural flexibility. Existing approximate or reconfigurable DCIM solutions tackle these issues partially without achieving a holistic balance. To address this, we propose PAR-CIM, a highly energy-efficient reconfigurable CIM macro that integrates precise and approximate paradigm. First, we introduce layer/gate-level approximate computation (LGAC) into the adder tree (AT) of the DCIM core, achieving full operation with only 0.35× the area of traditional implementations. Then, we develop a 0.35-4b fractional mixed-bitwidth quantization (FMBQ) algorithm, combining second-order Taylor sensitivity analysis with DoReFa-Net. This is complemented by a high-precision low-approximation (HPLA) mapping scheme to enhance energy efficiency. Additionally, a multi-bit reconfigurable computation mode (MBRM) strategy further improves architectural flexibility and enables the implementation of the proposed design. Under 40nm technology, PAR-CIM achieves 3048 TOPS/W at 1b/1b operations. With FMBQ, ResNet18 and our custom V-FuseMBA trained on CIFAR-10 achieve over 86.61% compression with accuracy loss under 0.74%, reaching classification accuracies of 93.67% and 92.86%, respectively.

Short video applications have attracted billions of users in recent years, fulfilling their various needs with diverse content. Users usually watch short videos on many topics on mobile devices in a short period of time, and give explicit or implicit feedback very quickly to the short videos they watch. The recommender system needs to perceive users' preferences in real-time in order to satisfy their changing interests. Traditionally, recommender systems deployed at server side return a ranked list of videos for each request from client. Thus it cannot adjust the recommendation results according to the user's real-time feedback before the next request. Due to client-server transmitting latency, it is also unable to make immediate use of users' real-time feedback. However, as users continue to watch videos and feedback, the changing context leads the ranking of the server-side recommendation system inaccurate. In this paper, we propose to deploy a short video recommendation framework on mobile devices to solve these problems. Specifically, we design and deploy a tiny on-device ranking model to enable real-time re-ranking of server-side recommendation results. We improve its prediction accuracy by exploiting users' real-time feedback of watched videos and client-specific real-time features.

In the new era of wireless communication, there has been an increased interest in indoor positioning systems that propelled researchers to come up with various solutions. Fusing the fuzzy locations from Bluetooth (BT) beacons with the ones from pedestrian dead reckoning (PDR) comes out a promising solution to provide meter-level positioning without additional infrastructure. Despite the remarkable efforts the community put to build the system, it lacks the performance examination of large-scale deployments. In this report, we implement and deploy a real system in Guangzhou South Railway Station to enable a large scale indoor positioning service. The framework used Bluetooth beacons and pedestrian dead reckoning to calculate the estimated position of the user then refine the accuracy through a fuzzy fusion algorithm. We distribute up to 2849 beacons in the indoor space of 81654 square meters. When applied to the real ceiling with a height of 8m and above, our approach is still able to reliably achieve a high accuracy of 5 meters. Also in buildings whose outer walls are nonstructural but large glass curtain wall that wraps around the roof of the facility, the GPS signal could serve as useful complement under careful utilization; last but not least significant, in practice we incorporate our indoor positioning services into WeChat HyperText Markup Language (HTML) platform, we discover that the headings provided by HTML5 are not always correct, the orientation error scale has a strong correlation with the temperature of phone, the residual error is accumulated along with the time to make matters worse.