About

Christopher Anderson is a lecturer in Film and Television at Boston University College of Communication. He is a musician, audio engineer, and sound designer with extensive experience working on acclaimed series such as Frontline, American Experience, and NOVA, as well as feature films including Detroit and American Hustle. His work also encompasses supervising sound editing on various features and shorts, and he has contributed to the Front Row Boston music series and several independent documentaries and short films. Anderson is on the staff at the Outpost at WGBH and runs his own business, Harpswell Sound Company, which specializes in independent feature audio consulting and sound supervision. His expertise emphasizes the importance of planning and knowledge in achieving good sound, and he enjoys exploring innovative ways to create unique sounds.

Research topics

• Computer science
• Political science
• Artificial intelligence
• Environmental science
• Machine learning

Selected publications

• A self-training spiking superconducting neuromorphic architecture

  npj Unconventional Computing · 2025-03-04 · 1 citations

  articleOpen accessSenior author

  Neuromorphic computing takes biological inspiration to the device level aiming to improve computational efficiency and capabilities. One of the major issues that arises is the training of neuromorphic hardware systems. Typically training algorithms require global information and are thus inefficient to implement directly in hardware. In this paper we describe a set of reinforcement learning based, local weight update rules and their implementation in superconducting hardware. Using SPICE circuit simulations, we implement a small-scale neural network with a learning time of order one nanosecond per update. This network can be trained to learn new functions simply by changing the target output for a given set of inputs, without the need for any external adjustments to the network. Further, this architecture does not require programing explicit weight values in the network, alleviating a critical challenge with analog hardware implementations of neural networks.

  PublisherOA PDFDOI

• Mapping potential habitat for naturally recolonizing cougars (Puma concolor) in Missouri, USA

  Global Ecology and Conservation · 2025-10-01

  articleOpen access

  Cougars (Puma concolor ) have been increasingly documented in the Midwestern United States in the past few decades following historical extirpation, including in Missouri (USA), where verified sightings occur regularly. To map potential habitat and inform monitoring efforts, we developed a species distribution model (MaxEnt) using a decade of verified cougar occurrences from 2011 to 2022 (n = 76) and landscape variables hypothesized to influence cougar occurrence. Model performance was good. Terrain ruggedness (50 %), distance to natural cover (23 %), and vegetation productivity (19 %) were the most influential predictors of landscape suitability for cougars. The model identified ∼63,000 km² of potential cougar habitat, primarily in central and southeastern Missouri; only 17 % of these overlapped with public lands. Model validation using 2023–2024 sightings (n = 19) showed 95 % of observations fell within or ≤ 3 km of potential habitat. Our model may slightly underestimate the full extent of cougar range, but it captures the overall spatial pattern and identifies a broader area of suitable habitat in Missouri than previous studies, highlighting the value of localized, state-level modeling for research and conservation planning. Findings provide a spatial framework for monitoring dispersing cougars in fragmented, human-modified landscapes, where public outreach and extension programs will be crucial to promoting effective human-cougar coexistence should recolonization occur in Missouri.

  PublisherDOI

• Self-training superconducting neuromorphic circuits using reinforcement learning rules

  arXiv (Cornell University) · 2024-04-29

  preprintOpen accessSenior author

  Reinforcement learning algorithms are used in a wide range of applications, from gaming and robotics to autonomous vehicles. In this paper we describe a set of reinforcement learning-based local weight update rules and their implementation in superconducting hardware. Using SPICE circuit simulations, we implement a small-scale neural network with a learning time of order one nanosecond. This network can be trained to learn new functions simply by changing the target output for a given set of inputs, without the need for any external adjustments to the network. In this implementation the weights are adjusted based on the current state of the overall network response and locally stored information about the previous action. This removes the need to program explicit weight values in these networks, which is one of the primary challenges that analog hardware implementations of neural networks face. The adjustment of weights is based on a global reinforcement signal that obviates the need for circuitry to back-propagate errors.

  PublisherOA PDFDOI

• SRViT: Vision Transformers for Estimating Radar Reflectivity from Satellite Observations at Scale

  arXiv (Cornell University) · 2024-06-20

  preprintOpen accessSenior author

  We introduce a transformer-based neural network to generate high-resolution (3km) synthetic radar reflectivity fields at scale from geostationary satellite imagery. This work aims to enhance short-term convective-scale forecasts of high-impact weather events and aid in data assimilation for numerical weather prediction over the United States. Compared to convolutional approaches, which have limited receptive fields, our results show improved sharpness and higher accuracy across various composite reflectivity thresholds. Additional case studies over specific atmospheric phenomena support our quantitative findings, while a novel attribution method is introduced to guide domain experts in understanding model outputs.

  PublisherOA PDFDOI

• Development of an affordable light emitting diode spectrophotometer paired with a Python program for calibration and linearity testing and the measurement of uranium(VI)

  PLoS ONE · 2024-09-17 · 2 citations

  articleOpen access

  Uranium (U) is a radiologically and chemically toxic element that occurs naturally in water, soil, and rock at generally low levels. However, anthropogenic uranium can also leach into groundwater sources due to mining, ore refining, and improper nuclear waste management. Over the last few decades, various methods for measuring uranium have emerged; however, most of these techniques require skilled scientists to run samples on expensive instrumentation for detection or require the pretreatment of samples in complex procedures. In this work, a Schiff base ligand (P1) is used to develop a simple spectrophotometric method for measuring the concentration of uranium (VI) with an accurate and affordable light-emitting diode (LED) spectrophotometer. A test for a higher-order polynomial relationship was used to objectively determine the calibration data's linearity. This test was done with a Python program on a Raspberry Pi computer that captured the spectrophotometer's calibration and sample measurement data.

  PublisherDOI

• A harvest framework for a recovering American black bear population

  Journal of Wildlife Management · 2023-10-10 · 1 citations

  articleOpen access

  Abstract Having reproducible and transparent science‐based processes in wildlife management ensures the integrity of decision making. These processes are particularly important when establishing harvest frameworks, as guiding information in the peer‐reviewed literature is limited. We provide an example using multiple data sets, whose products guided aspects of the development of a harvest framework for a population of recolonizing American black bears ( Ursus americanus ) in Missouri, USA. To characterize the spatial distribution of harvest, we used 10 years (2010–2019) of black bear global positioning system (GPS) location data and 30 years (1991–2020) of sightings data to assess spatial vulnerability to harvest as the intersection among information on bear occurrence, bear sightings, and hunter land‐use tendencies (i.e., the avoidance of steep slopes, large distances from roads). We then used the spatial vulnerability assessment, information on the distribution of public and private lands, and easily discernable boundaries (i.e., major highways, rivers) to suggest boundaries for bear management zones. Additionally, to identify the timing of harvest that would limit female harvest bias, we assessed the temporal vulnerability of harvest using sex‐based changes in average daily step lengths and monthly utilization distribution sizes during fall. Black bear occurrence and sighting propensity was greater in southwestern Missouri, and potential hunter land use appeared pervasive across the landscape given the lack of landscape features that would disincentivize use. Given the influence of black bear occurrence and sighting propensity, spatial harvest vulnerability diminished from southern and southeastern to central portions of Missouri, with areas north of the Missouri River not a part of the established black bear range. We consequently divided areas south of the Missouri River into 3 black bear management zones: a small southwestern zone with primarily private lands and high harvest vulnerability, a southeastern zone that encompassed considerable public lands and moderate amounts of vulnerability, and a central zone that was composed mainly of areas of low vulnerability. Temporally, males did not exhibit movement‐based changes, but females became less active after the first week of October and used 63.9% less area through fall. Based on movements rates of males and females, a hunting season after the first week of October could reduce the likelihood of females being harvested. Harvests from the black bear harvest season in 2021 suggest that the proportion of bears harvested in each zone was similar in distribution to the proportion of permits allocated across zones with no harvest sex bias, which was aligned with agency goals. Animal movement and space use data products can guide harvest framework decision‐making.

  PublisherOA PDFDOI

• Exploring the Use of Machine Learning to Improve Vertical Profiles of Temperature and Moisture

  Artificial Intelligence for the Earth Systems · 2023-10-27 · 5 citations

  articleOpen access

  Abstract Vertical profiles of temperature and dewpoint are useful in predicting deep convection that leads to severe weather, which threatens property and lives. Currently, forecasters rely on observations from radiosonde launches and numerical weather prediction (NWP) models. Radiosonde observations are, however, temporally and spatially sparse, and NWP models contain inherent errors that influence short-term predictions of high impact events. This work explores using machine learning (ML) to postprocess NWP model forecasts, combining them with satellite data to improve vertical profiles of temperature and dewpoint. We focus on different ML architectures, loss functions, and input features to optimize predictions. Because we are predicting vertical profiles at 256 levels in the atmosphere, this work provides a unique perspective at using ML for 1D tasks. Compared to baseline profiles from the Rapid Refresh (RAP), ML predictions offer the largest improvement for dewpoint, particularly in the middle and upper atmosphere. Temperature improvements are modest, but CAPE values are improved by up to 40%. Feature importance analyses indicate that the ML models are primarily improving incoming RAP biases. While additional model and satellite data offer some improvement to the predictions, architecture choice is more important than feature selection in fine-tuning the results. Our proposed deep residual U-Net performs the best by leveraging spatial context from the input RAP profiles; however, the results are remarkably robust across model architecture. Further, uncertainty estimates for every level are well calibrated and can provide useful information to forecasters.

  PublisherOA PDFDOI

• Workshops of the eighth international brain–computer interface meeting: BCIs: the next frontier

  Brain-Computer Interfaces · 2022-02-08 · 11 citations

  articleOpen access

  The Eighth International Brain-Computer Interface (BCI) Meeting was held June 7-9th, 2021 in a virtual format. The conference continued the BCI Meeting series' interactive nature with 21 workshops covering topics in BCI (also called brain-machine interface) research. As in the past, workshops covered the breadth of topics in BCI. Some workshops provided detailed examinations of specific methods, hardware, or processes. Others focused on specific BCI applications or user groups. Several workshops continued consensus building efforts designed to create BCI standards and increase the ease of comparisons between studies and the potential for meta-analysis and large multi-site clinical trials. Ethical and translational considerations were both the primary topic for some workshops or an important secondary consideration for others. The range of BCI applications continues to expand, with more workshops focusing on approaches that can extend beyond the needs of those with physical impairments. This paper summarizes each workshop, provides background information and references for further study, presents an overview of the discussion topics, and describes the conclusion, challenges, or initiatives that resulted from the interactions and discussion at the workshop.

  PublisherOA PDFDOI

• Spatially-explicit population modeling to predict large carnivore recovery and expansion

  Ecological Modelling · 2022-05-24 · 4 citations

  articleOpen access

  Despite extensive range contractions, several large carnivore species have recently recolonized portions of their former range. Since the 1990s, American black bear reports have increased in Missouri, USA, corresponding with increasing abundance and distribution. As effective management benefits from sound information on wildlife demographics and spatial distribution, we used a spatially-explicit metapopulation model built from local data to inform a stage matrix, which was used in conjunction with a previously created habitat suitability model to quantify and predict the growth and expansion of the black bear population. We collected demographic data during 2011–2021 from 162 female bears with 159 young from 70 litters. Average litter size was 1.95, 42% of cubs were female, and cub survival to yearling was, on average, 95% for females and 70% for males. The habitat suitability model identified 53 core areas most suitable for bears in Missouri. Increasing from previous local estimates of bear population size (∼280 bears in 2012), the model estimated the total bear population inhabiting core habitat patches in year 10 as 732 individuals (598–873, 5th–95th percentiles), with an average annual growth rate of 1.11. The estimated carrying capacity of the core habitat patches (excluding cubs-of-the-year) was ∼2700 bears, but adding marginal and highly marginal habitat patches increased the carrying capacity to ∼5500 bears. Population estimates were most sensitive to variations in stage matrix parameters though overall patch colonization estimates were relatively stable. Our work is an important step toward understanding the recolonization potential of black bears in Missouri and can aid future studies projecting large carnivore density in other areas as well as values and attitudes toward these wide-ranging species. Developing holistic conservation frameworks will be critical if human acceptance toward large carnivores continues increasing and carnivore recolonization of historical range continues.

  PublisherDOI

• Abstract 9733: Echocardiography Workshops Are an Effective Learning Tool for Pre-Clerkship Medical Students

  Circulation · 2022-11-08

  articleSenior author

  Introduction: Point of care ultrasound curriculum is being taught in undergraduate medical education (UME) with greater frequency and demonstrable benefit to students’ learning. While echocardiography (echo) curricula have been implemented in a small number of medical schools, there is limited data on the efficacy of the learning experience. Hypothesis: Echo workshops in UME are feasible and effective as a means of introducing medical students to basic echo skills. Methods: 24 pre-clerkship medical students volunteered to undergo 4-hour echo workshops. A variety of preparatory learning materials covering basic aspects of echo were provided to students one week prior to the workshop. A short quiz was administered to all students immediately before and after the workshop. Quiz questions were coded as cardiac anatomy (6 questions), basic echo imaging (8 questions), basic ultrasound and Doppler theory (2 questions), and advanced echo imaging techniques (3 questions). During the workshop, each student was provided at least 30 minutes of hands-on imaging while another student acted as the “patient.” Certified echosonographers and pediatric cardiologists served as the instructors during the workshop. Students completed a survey to characterize the quality of the learning experience. Students’ performance on the quiz pre- and post-workshop were compared using student’s t-test. Results: The mean overall pre- and post-workshop quiz scores were 6.8/19 and 11.4/19, respectively (p<0.00001). Mean pre- vs post-quiz scores for cardiac anatomy were 2.0 vs 3.8/6 (p=0.0001); for basic echo imaging were 3.5 vs 5.3/8 (p<0.001); for basic ultrasound and Doppler theory were 0.5 vs 1.5/2 (p<0.00001); and for advanced echo imaging were 0.7 vs 0.9/3 (p=0.5), respectively. All students characterized the quality of their learning experience through the workshop as a very effective use of their time. Conclusions: Echo workshops targeted to pre-clerkship medical students are feasible, and provide effective learning experiences in cardiac anatomy, basic echo imaging, and basic ultrasound and Doppler theory in the short-term, and are very well-received by students. Providing basic echo curriculum in the pre-clerkship years of medical school should be further explored.

  PublisherDOI

Recent grants

• HCC: Medium: Removing Barriers to the Practical Use of Non-Invasive Brain-Computer Interfaces

  NSF · $1.2M · 2011–2018

• Geometric Pattern Analysis and Mental Task Design for a Brain-Computer Interface

  NSF · $834k · 2002–2012

• Alternate Modes of Human-Computer Interaction: EEG Recognition with Neural Networks

  NSF · $285k · 1992–1996

Frequent coauthors

• Shlomo Avineri

  3535 shared

• Frederick Crosson

  3485 shared

• Gerald Garvey

  Bowdoin College

  3455 shared

• Philip Gleason

  3396 shared

• Ernest L. Fortin

  3391 shared

• Arend Lijphart

  3346 shared

• Glenn Tinder

  3334 shared

• Donald P. Kommers

  3227 shared

Education

• Ph.D., Computer Science

  University of Massachusetts Amherst

  1986

• M.S., Computer Science

  University of Massachusetts Amherst

  1982

• B.S., Computer Science

  University of Nebraska-Lincoln

  1978